Skip to main content

Full text of "The journal of raptor research"

See other formats 






THE RAPTOR RESEARCH FOUNDATION, INC. 

(Founded 1966) 

http: //biology.boisestate.edu/ raptor/ 

OFFICERS 

SECRETARY: Judith Henckei. 
TREASURER: Jim Fitzpatrick 

BOARD OF DIRECTORS 

INTERNATIONAL DIRECTOR #3: 

Steve Redpatii 

DIRECTOR AT LARGE #1: Jemima Parry-Jones 
DIRECTOR AT LARGE #2: Eduardo Inigo-Elias 
DIRECTOR AT LARGE #3: Michael W. Collopy 
DIRECTOR AT LARGE #4: Carol McIntyre 
DIRECTOR AT LARGE #5: John A. Smallwood 
DIRECTOR AT LARGE #6: Daniel E. Varland 

Ruth Tingay 


NORTH AMERICAN DIRECTOR #1: 
Steve Hoffman 

NORTH AMERICAN DIRECTOR #2: 
Gary Santolo 

NORTH AMERICAN DIRECTOR #3: 
Ted Swem 

INTERNATIONAL DIRECTOR#!: 
Nick Mooney 

INTERNATIONAL DIRECTOR #2: 


PRESIDENT: Brian A. Millsap 
VICE-PRESIDENT: David M. Bird 


EDITORIAL STAFF 


EDITOR: James C. Bednarz, Department of Biological Sciences, RO. Box 599, Arkansas State 
University, State University, AR 72467 U.S.A. 


ASSOCIATE EDITORS 

James R. Belthofe Joan L. Morrison 

Clint W. Boat Fabrizio Sergio 

Chfryi. R. Dykstra Ian G. Warkentin 

Michael I. Goi dstein James W. Watson 

BOOK REVIEW EDITOR: Jeffrey S. Marks, Montana Cooperative Research Unit, University of Montana, 

Missoula, MT 59812 U.S.A. 

SPANISH EDITORS: Carlos Daniel Cadena, Lucio R. Malizia, Cintia Cornelius 
EDITORIAL ASSISTANTS: Autumn Earless, Joan Clark 

The Journal of Raptor Research is distributed quarterly to all current members. Original manuscripts 
dealing with the biology and conservation of diurnal and nocturnal birds of prey are welcomed from 
throughout the world, but must be written in English. Submissions can be in the form of research articles, 
short communications, letters to the editor, and book reviews. Contributors should submit a typewritten 
original and three copies to the Editor. All submissions must be typewritten and double-spaced on one 
side of 216 X 278 mm {SVi X 11 in.) or standard international, white, bond paper, with 25 mm (1 in.) mar- 
gins. The cover page should contain a title, the author’s full name(s) and address (es). Name and address 
should be centered on the cover page. If the current address is different, indicate this via a footnote. A 
short version of the title, not exceeding 35 characters, should be provided for a running head. An 
abstract of about 250 words should accompany all research articles on a separate page. 

Tables, one to a page, should be double-spaced throughout and be assigned consecutive Arabic numer- 
als. Collect all figure legends on a separate page. Each illustration should be centered on a single page 
and be no smaller than final size and no larger than twice final size. The name of the author(s) and figure 
number, assigned consecutively using Arabic numerals, should be pencilled on the back of each figure. 

Names for birds should follow the A.O.U. Checklist of North American Birds (7th ed., 1998) or another 
authoritative source for other regions. Subspecific identification should be cited only when pertinent to 
the material presented. Metric units should be used for all measurements. Use the 24-hour clock (e.g., 
0830 H and 2030 H) and “continental” dating (e.g., 1 January 1999). 

Refer to a recent issue of the journal for details in format. Explicit instructions and publication policy 
are outlined in “Information for contributors,”/. Raptor Res., Vol. 38(4), and are available from the editor. 
Submit manuscripts to J. Bednarz at the address listed above. 


COVER; Northern Goshawk {Accipiter gentilis) . Painting done in gouache by Michael Demain; for more 
information and images, visit www.michaeldemainwildlifeart.co.uk 


Contents 


Introduction 

Preface: Proceedings oe the International Symposium on the Ecology and Management of 

Northern Goshawks. Clint W. Boal 189 

In Memoriam: Suzanne MeridethJoy. Richard T. Reynolds 190 

Status 

Technical Review of the Status of Northern Goshawks in the Western United States. 

David E. Andersen, Stephen DeStefano, Michael I. Goldstein, Kimberly Titus, Cole 
Crocker-Bedford, John J. Keane, Robert G. Anthony, and Robert N. Rosenfield 192 

Biology 

Is Fledging Success a Reliable Index of Fitness in Northern Goshawks? 

J. David Wiens and Richard T. Reynolds 210 

Productivity and Mortality of Northern Goshawks in Minnesota. 

Clint W. Boal, David E. Andersen, and Patricia L. Kennedy 222 

Relationships Between Winter and Spring Weather and Northern Goshawk (Accipiter gentilis) 

Reproduction in Northern Nevada. Graham D. Fairhurst and Marc J. Bechard 229 

Patterns of Temporal Variation in Goshawk Reproduction and Prey Resources. 

Susan R. Salafsky, Richard T. Reynolds, and Barry R. Noon 237 

A Skewed Sex Ratio in Northern Goshawks: Is It a Sign of a Stressed Population? 

Michael E Ingraldi 247 

Northern Goshawk {Accipiter gentilis iaingi) Post-fledging Areas on Vancouver Island, British 

Columbia. Erica L. McClaren, Patricia L. Kennedy, and Donald D. Doyle 253 

Northern Goshawk Diet in Minnesota: An Analysis Using Video Recording 

Systems. Brett L. Smithers, Clint W. Boal, and David A. Andersen 264 

Techniques 

Sampling Considerations for Demographic and Habitat Studies of Northern Goshawks. 

Richard T. Reynolds, J. David Wiens, Suzanne M. Joy, and Susan R. Salafsky 274 

Population Genetics and Genotyping for Mark-Recapture Studies oe Northern Goshawks 
{Accipiter gentius) on the Kaibab Plateau, Arizona. Shelley Bayard de Volo, Richard T. 

Reynolds, J. Rick Topinka, Bernie May, and Michael F. Antolin 286 

When Are Goshawks Not There? Is a Single Visit Enough to Infer Absence at Occupied Nest 
Areas? Douglas A. Boyce, Jr., Patricia L. Kennedy, Paul Beier, Michael F. Ingraldi, 

Susie R. MacVean, Melissa S. Siders, John R. Squires, and Brian Woodbridge 296 

Quantifying Northern Gosfiawk Diets Using Remote Cameras and Observations 

FROM Blinds. Audi S. Rogers, Stephen DeStefano, and Micheal F. Ingraldi 303 

Conservation 

Temporal Patterns of Northern Goshawk Nest Area Occupancy and Habitat: A Retrospective 

Analysis. Steven Desimone and Stephen DeStefano 310 

Monitoring Results of Northern Goshawk Nesting Areas in the Greater Yellowstone Ecosystem: 

Is Decline in Occupancy Related to Habitat Change? Susan M. Pada 324 

Effects of Timber Harvesting Near Nest Sites on the Reproductive Success of Northern 

Goshawks {Accipiter gentius). Todd Mahon and Frank I. Doyle 335 

A Review of the Status and Distribution of Northern Goshawks in New Engiand. 

Stephen DeStefano 342 


The Raptor Research Foundation, Inc. gratefully acknowledges funds and logistical support 
provided by Arkansas State University to assist in the publication of the journal. 


THE JOURNAL OF RAPTOR RESEARCH 

A QUARTERLY PUBLICATION OF THE RAPTOR RESEARCH FOUNDATION, INC. 


VoL. 39 September 2005 No. 3 


J Raptor Res. 39(3):189 

© 2005 The Raptor Research Foundation, Inc. 


PREFACE 

PROCEEDINGS OF THE INTERNATIONAL SYMPOSIUM ON THE 
ECOLOGY AND MANAGEMENT OF NORTHERN GOSHAWKS 

Clint W. Boal^ 

U.S. Geological Survey, Texas Cooperative Fish and Wildlife Research Unit, Department of Range, Wildlife, and Fisheries 

Management, Texas Tech University, Lubbock, TX 79409-2120 U.S. A. 


For almost 20 years, the Northern Goshawk {Ac- 
cipiter gentilis) has been the focus of considerable 
research effort, management and conservation 
planning, and litigation. This has been driven by 
conflict between conservation concerns and forest 
management practices. Due to increasing concern 
for the species, the symposium “The Biology and 
Management of Northern Goshawks,” was held in 
conjunction with the Cooper Ornithological Soci- 
ety in 1993, with the goal of assembling the most 
current information available at that time. 

Since that 1993 symposia, 10 years had passed 
during which a considerable amount of research 
had been conducted on goshawks, but a venue for 
sharing the information among researchers had not 
been made available. To provide an opportunity for 
researchers and managers to exchange information 
with which to assess the current state of knowledge 
on Northern Goshawks, a coordinating committee 
consisting of Steve DeStefano, Patricia Kennedy, Mi- 
chael Goldstein, John Keane, and myself organized 
the “International Symposium on the Ecology and 
Management of Northern Goshawks.” This sympo- 
sium was held 4—5 September 2003, in conjunction 
with the 2003 Raptor Research Foundation Annual 
Meeting in Anchorage, Alaska. The symposium was 
well attended, with 32 papers presented. Topics in- 
cluded status reports; improvements in survey and 
monitoring methodology; information on popula- 
tion demography, food habits, habitat assessment, 
and winter ecology; and the use of genetic ap- 
proaches to goshawk studies. 

Given the obvious interest in goshawk ecology 


^ E-mail address: cboal@ttu.edu 


and management, as indicated by attendance and 
number of papers presented at the symposium, 
there was a need to make this information more 
broadly available. We have attempted to do so with 
this issue of The Journal of Raptor Research, which 
includes peer-reviewed versions of some of the pa- 
pers presented at the 2003 symposia. The lead pa- 
per is authored by a committee put together jointly 
by The Raptor Research Foundation and The 
Wildlife Society. This committee was charged with 
reviewing the status of Northern Goshawks in the 
western United States. This review provides a con- 
text for the rest of this issue in presenting an over- 
view of the state of knowledge leading up to the 
symposium. Manuscripts following the lead paper 
are grouped topologically (i.e.. Biology, Tech- 
niques, and Conservation). 

All manuscripts in this issue went through the 
same peer-review process as required for regular 
issues of the Journal. For this issue, I served as a 
Coordinating Editor, with the symposia committee 
members serving as Associate Editors. I, the Asso- 
ciate Editors, and Jim Bednarz (Editor-in-Chief) , 
would like to thank all of the individuals that 
served as peer-referees for the manuscripts in this 
issue. We would also like to acknowledge the U.S. 
Geological Survey Cooperative Research Units, the 
North American Falconer’s Association, and the 
U.S. Forest Service Alaska Region and the Chu- 
gach National Forest, for providing support for 
publication of this issue of the Journal. 

Finally, this issue of The Journal of Raptor Research 
is dedicated to the memory of Suzanne Joy, a 
friend and colleague to many of us working with 
goshawks. Please see the memoriam provided by 
Richard Reynolds on the following page. 


189 


J. Raptor Res. 39 (3); 190-1 91 
© 2005 The Raptor Research Foundation, Inc. 


IN MEMORIAM 
Suzanne Merideth Joy 
12 APRIL 1961-7 DECEMBER 2004 

Richard T. Reynolds 

Rocky Mountain Research Station, USDA Forest Service, 2150 Centre Avenue, Building A, Suite 350, 

Fort Collins, CO 80526-1891 U.S.A. 

Suzanne Merideth Joy, microbiologist, photographer, animal ecologist, raptor biologist, tree climber extraordinaire, 
spatial analyst, and always a mentor, passed away on 7 December 2004 in Teramo, Italy. She was in Italy developing 
spatial models of the habitat of biting midge (Ceratopogomidae) vectors of bluetongue disease (Reoviridae) in cattle 
and sheep on behalf of the Italian government and the U.S. Department of Agriculture Animal and Plant Health 
Inspection Service (APHIS), her employer since 2003. Suzanne was born in Goose Bay, Labrador, but grew up in 
Virginia, Madrid (Spain), Minnesota, and Arizona. After graduating from Colorado State University with honors (Phi 
Beta Kappa, Phi Kappa Phi) and as the outstanding senior in Microbiology in 1983, she spent 2 yr with the Peace 
Corps in Kenya, East Africa. She was conversant in French, Swahili, and Kikuyu. Upon returning from Kenya, she 
worked as a radioimmunoassay technician for Hazleton Biotechnologies Company in Massachusetts and researched 
a pertussis vaccine for the Food and Drug Administration in Bethesda, Maryland. 

In 1987, wanting to move from the confines of research labs to the natural habitats of birds and mammals, Suzanne 
returned to Colorado State University for a Master of Science (1990) degree in Fishery and Wildlife Biology. Her 
thesis topic was the feeding ecology and nest habitat of the Sharp-shinned Hawk {Acdpiter striatus) in Colorado’s 
Rocky Mountain forests. The experience and knowledge she gained while sampling for nesting Acdpiter, collecting 
and identifying their food remains, and measuring the composition and structure of their forest nest sites was soon 
put to use in helping me establish in 1991 what would turn out to be an intensive and long-term study of the 
demography, genetics, and habitat of Northern Goshawks (A. gentilis) on the Kaibab Plateau, AZ. Before beginning 
a Ph.D. program at Colorado State University, Suzanne helped develop protocols for finding and trapping goshawks 
and managing budgets and large field crews through much of the Kaibab goshawk study. She loved being in the 
piney woods searching for and trapping goshawks, and making difficult tree climbs to goshawk nests to band their 
young. In 2002, Suzanne was awarded her Ph.D. by the Graduate Degree Program in Ecology at Colorado State 
University. Her dissertation, entitled “Northern Goshawk Habitat on the Kaibab National Forest in Arizona: Factors 
Affecting Nest Locations and Territory Quality,” included a dynamic spatial simulation model that described the 
spatial dependence of goshawk nest locations on both territoriality and the availability of suitable nest sites. She 
identified the correlates of habitat quality by quantifying the relationship between the long-term reproductive per- 
formance of breeding goshawks and the composition and structure of habitats within their territories. Within a few 
months of completing her dissertation, Suzanne moved to APHIS, where she worked as a spatial analyst, and where 
once again she quickly became a star. 

Suzanne Joy is survived by a son, Quinn, who shows his mother’s love of nature, and husband, Vern Thomas, both 
of whom brought much additional joy to her life. Suzanne will be greatly missed by all those whose lives she touched. 
I was privileged to have known Suzanne as a student, employee, colleague, mentor, and above all, a friend. . 

The editors and authors of these proceedings dedicate the following contributions on the biology and conservation 
of the Northern Goshawk to Suzanne Joy’s memory. 


190 


September 2005 


In Memoriam 


191 


IN MEMORIAM 
Suzanne Merideth Joy 
12 APRIL 1961-7 DECEMBER 2004 



Suzanne Joy with captured Northern Goshawk in the Kaibab National Forest, Arizona. 



J. Raptor Res. 39(3):192-209 
© 2005 The Raptor Research Foundation, Inc. 


TECHNICAL REVIEW OF THE STATUS OF NORTHERN 
GOSHAWKS IN THE WESTERN UNITED STATES^ 

David E. Andersen^ 

U.S. Geological Survey, Minnesota Cooperative Fish and Wildlife Research Unit, St. Paul, MN 35108 U.S.A. 

Stephen DeStefano 

U.S. Geological Survey, Massachusetts Cooperative Fish and Wildlife Research Unit, Amherst, MA 01003 U.S.A. 

Michael L Goldstein 

U.S. Forest Service, Chugach National Forest, Anchorage, AK 99503 U.S.A. 

Kimberly Titus 

Alaska Department of Fish and Game, funeau, AK 99802 U.S.A. 

Cole Crocker-Bedford 

U.S. Park Service, Grand Canyon National Park, Grand Canyon, AZ 86023 U.S.A. 

JohnJ. Keane 

US. Forest Service, Pacific Southwest Research Station, Sierra Nevada Research Center, Davis, CA 95616 U.S.A. 

Robert G. Anthony 

U.S. Geolo^cal Survey, Oregon Cooperative Fish and Wildlife Research Unit, Corvallis, OR 97331 U.S.A. 

Robert N. Rosenfield 

Department of Biology, University of Wisconsin— Stevens Point, Stevens Point, WI 54481 U.S.A. 

Abstract. — The U.S. Fish and Wildlife Service (FWS) was petitioned in 1997 to consider listing North- 
ern Goshawks (Accipiter gentilis atricapillus) under the Endangered Species Act of 1973, west of the 100* 
meridian of the contiguous United States. In their 12-mo finding issued in June 1998, the FWS deter- 
mined that listing this population as threatened or endangered was not warranted and based that 
decision on review of existing population and habitat information. Because the status of goshawks in 
the western U.S. continues to be contentious and the FWS finding has been challenged, the Raptor 
Research Foundation, Inc. and The Wildlife Society jointly formed a committee to review information 
regarding the status of the goshawk population in the contiguous U.S. west of the 100* meridian. The 
committee was requested to: (1) determine if there is evidence of a population trend in goshawks in 
the western U.S., excluding Alaska; (2) determine if there is evidence that goshawks nesting in the 
eastern and western U.S. represent distinctive, genetically unique populations; and (3) evaluate evidence 
for goshawk— habitat relations, including any association with large, mostly-unbroken tracts of old growth 
and mature forests. Based on existing information, the committee concluded: (1) existing data are not 
adequate to assess population trend in goshawks west of the 100* meridian; (2) existing analyses of 
phylogeography have not provided evidence of genetic differences among recognized {atricapillus, 
laingi) or putative {apache) subspecies, and the genetic distinctness of atricapillus goshawks in western 
and eastern North America is not known; and (3) at present, assessing the status of goshawks solely 
using distribution of late-successional forests is not appropriate, based on the current understanding of 
goshawk-habitat relations, although goshawks clearly use and often select late-successional forests for 


* Summary of a report prepared by the Technical Committee on the Status of Northern Goshawks in the Western 
United States sponsored jointly by the Raptor Research Foundation, Inc. and The Wildlife Society. A copy of the 
complete report can be obtained from The Wildlife Society: (http://www.wildlife.org). 

^ Corresponding author’s email address: dea@umn.edu 


192 


September 2005 


Status 


193 


nesting and foraging. We provide recommendations on information needs to assess status and popula- 
tion trend of goshawks in the western U.S. 

Keywords: Northern Goshawk, Accipiter gentilis atricapillus; western U.S.; status', review, population trend', 

genetic structure, habitat relations. 


REVISION TECNICA DEL ESTATUS DE ACCIPITER GENTILIS ATRICAPILLUS EN EL OESTE DE LOS 
ESTADOS UNIDOS 

Resumen. — El Servicio de Pesca y Vida Silvestre de los Estados Unidos (FWS, por sus siglas en ingles) 
recibio en 1997 la peticion de considerar a Accipiter gentilis atricapillus al oeste del meridiano 100 de los 
Estados Lfnidos (considerando solo los estados contiguos) como un ave amenazada de acuerdo al acta 
de 1973. Luego de 12 meses, en junio de 1998 el FWS dictamino que clasificar a esta poblacion como 
amenazada o en peligro no era justificable, y baso dicha decision en una revision de la informacion 
poblacional y de habitat existente. Debido a que el estatus de A. g. atricapillus en el oeste de los Estados 
Unidos es aun controversial y a que el hallazgo del FWS ha sido desafiado, la Raptor Research Foun- 
dation, Inc. y la Wildlife Society formaron un comite conjunto para revisar la informacion concerniente 
al estatus de la poblacion de esta ave en los estados contiguos de los Estados Unidos al oeste del 
meridiano 100. Al comite se le solicito que (1) determinara si existe evidencia de una tendencia de 
cambio en el tamaho poblacional de A. g. atricapillus en el oeste de E. U., excluyendo Alaska; (2) 
determinara si existe evidencia de que los individuos que nidifican en el este y el oeste de E. U. 
representan poblaciones distintivas, geneticamente unicas; y (3) evaluara la evidencia sobre las rela- 
ciones de A. g. atricapillus con el habitat, incluyendo cualquier asociacion con reductos grandes y no 
fragmentados de bosques maduros. Con base en la informacion existente, el comite concluyo que: (1) 
los datos disponibles son inadecuados para evaluar si existe una tendencia de cambio en el tamaho 
poblacional al oeste del meridiano 100; (2) los analisis de filogeografia existentes no han provisto 
evidencia que indique la existencia de diferencias entre las subespecies reconocidas {atricapillus, laingi) 
o putativas {apache), y no se conocen diferencias geneticas entre las poblaciones del oeste y el este de 
Norte America; y (3) en la actualidad, evaluar el estatus de A. g. atricapillus con base solo en la distri- 
bucion de bosques de estadios sucesionales tardios no es adecuado de acuerdo al conocimiento actual 
de sus relaciones con el habitat, aunque es claro que esta ave utiliza y a menudo selecciona bosques de 
sucesion avanzada para nidificar y forrayear. Ofrecemos recomendaciones en cuanto a la informacion 
necesaria para evaluar el estatus y las tendencias poblacionales de A. g. atricapillus en el oeste de Estados 
Unidos. 

[Traduccion del equipo editorial] 


In 1997, the U.S. Fish and Wildlife Service 
(FWS) received a petition to list the Northern Gos- 
hawk {Accipiter gentilis atricapillus; hereafter re- 
ferred to as goshawk) west of the 100‘^ meridian 
of the contiguous United States under the Endan- 
gered Species Act of 1973. In its 90-d finding issued 
in September 1997 (United States Department of 
Interior [USDI] 1997), the FWS found that the pe- 
tition “presented substantial information indicat- 
ing that the listing of the Northern Goshawk as a 
threatened or endangered species in the contigu- 
ous United States west of the 100* meridian may 
be warranted” (USDI 1998). The FWS at that time 
initiated a status review (FWS 1998) for the gos- 
hawk, and in June 1998 issued its 12-mo petition 
finding (USDI 1998) and indicated that after “. . . 
reviewing all available scientific and commercial in- 
formation, the Service finds that listing this popu- 


lation as endangered or threatened is not warrant- 
ed” (USDI 1998:35183). 

The FWS used data from recent survey and mon- 
itoring that suggested goshawks had generally been 
located where intensive survey and monitoring ef- 
forts were implemented, and that goshawks re- 
mained widely distributed throughout their histor- 
ical range. The FWS also reviewed existing habitat 
data and concluded that there was no evidence 
that habitat was currently limiting the goshawk 
population in the western U.S., and habitat was un- 
likely to limit this population in the foreseeable 
future. The petition for listing suggested that gos- 
hawks in the western U.S. were dependent upon 
large, unbroken tracts of late-successional forest, 
but the FWS concluded that there was little or no 
support for that assertion. Subsequent to release of 
the 12-mo finding by the FWS, several court chal- 


194 


Andersen et al. 


VoL. 39, No. 3 


lenges were submitted, both to the finding itself 
and to the process used to arrive at the finding. 

Clearly, there is considerable concern for con- 
servation of goshawk populations and their habi- 
tats in western North America. As some of the fore- 
most professional societies concerned with 
conservation of wildlife in general, and raptors in 
particular, the Raptor Research Foundation, Inc. 
(RRF) and The Wildlife Society (TWS) formed a 
joint committee to review information regarding 
the status of the goshawk, population in the western 
contiguous U.S. The purpose behind forming this 
committee was to provide an independent techni- 
cal review of existing information related to gos- 
hawk status and to identify additional information 
necessary to assess population trend adequately. 
Specifically, the committee was requested to: (1) 
determine if there was evidence of a population 
trend in goshawks in the western U.S. west of the 
100* meridian, excluding Alaska; (2) determine if 
there was evidence that goshawks nesting in the 
eastern and western U.S. represent genetically dis- 
tinct populations; and (3) evaluate evidence for 
goshawk-habitat relations, including any associa- 
tion with large, mostly unbroken tracts of old- 
growth and mature forests. In addition, the com- 
mittee was asked to evaluate existing information 
on population trend, genetic structure, and habitat 
relations and to identify types of information need- 
ed to assess the status of goshawks more conclu- 
sively in the western U.S., excluding Alaska. This 
manuscript summarizes the process used, infor- 
mation evaluated, and opinions of the Joint RRF- 
TWS Technical Committee on the status of North- 
ern Goshawks in the western United States. A copy 
of the complete report can be obtained from TWS 
(http://www.wildlife.org) . 

Methods 

The scope of the committee’s review and evaluation 
was restricted to pertinent technical information, com- 
prised of peer-reviewed primary literature, theses, or un- 
published technical information that the committee 
deemed credible and that related directly to the com- 
mittee’s charge. Information considered included that 
summarized in the FWS goshawk status review (USDI 
1998) and related documents (e.g., FWS 1998), syntheses 
of the published literature (e.g., Squires and Reynolds 
1997), and published and unpublished information not 
included in previous reviews and available to the com- 
mittee through completion of its charge in 2003. Where 
possible, the committee reviewed primary literature and 
data, rather than relying solely on published or unpub- 
lished syntheses. Committee deliberations focused on 
three mzyor areas: (1) population trends, (2) genetic 


structure, and (3) goshawk— habitat associations. In ad- 
dition, as a fourth area, the committee considered recent 
conservation efforts that focused on the possibility of us- 
ing goshawk-habitat associations and habitat monitoring 
as a surrogate for population monitoring. 

Results 

Population Trends. Migration counts. Migration 
counts have several mEyor drawbacks as an index 
to the population size of goshawks in western 
North America. First, there is a nearly complete 
lack of knowledge of the geographic origin (e.g., 
breeding grounds) of goshawks observed at count 
locations. Second, migration routes for goshawks 
in western North America are poorly known 
(Squires and Reynolds 1997). Third, a primary lim- 
itation of migration counts is that changes in 
counts (FWS 1998) have an unknown relation to 
changes in the size of the target population (Ken- 
nedy 1998). Fourth, many migration counting sta- 
tions, especially in western North America (FWS 
1998), have small counts of migrating goshawks. 
Fifth, counting effort at some migration sites is var- 
iable through time and would need to be stan- 
dardized if counts were to be used as an index to 
population size (Mueller et al. 1977, Bednarz et al. 
1990, Bildstein 1998). Finally, continental counts 
included in the FWS status review (FWS 1998) are 
comprised primarily of counts of migrating gos- 
hawks from a single, more eastern site — Hawk 
Ridge near the western end of Lake Superior. For 
these reasons, migration counts at present are not 
a reliable index of goshawk population size in west- 
ern North America. 

Trend data. Breeding Bird Survey (BBS) data are 
inadequate to estimate population trends for gos- 
hawks across the western U.S., both because the 
number of routes on which goshawks were detect- 
ed (<35) and the encounter rate of goshawks on 
these routes (mean detection rate <0.02 goshawks 
per route) were too low. Christmas Bird Count 
(CBC) data were also inadequate to estimate gos- 
hawk population trends at large scales because of 
low encounter rates. In addition, the CBC is con- 
ducted outside of the breeding season, thereby 
making the origin of observed birds uncertain. 
Thus, observed trends in CBC data cannot be re- 
lated to the population of goshawks breeding in 
the western U.S. 

Productivity. Existing data on goshawk reproduc- 
tion in the western U.S. suggest that annual pro- 
ductivity (e.g., FWS 1998, Ingraldi 1998, Reynolds 
and Joy 1998) and nest success (Squires and Reyn- 


September 2005 


Status 


195 


olds 1997, FWS 1998, Ingraldi 1998, Reynolds and 
Joy 1998) are highly variable. Interpretation of 
studies of goshawk productivity is further con- 
founded by small sample sizes (e.g., FWS 1998) 
and biases in estimates of breeding area occupancy 
and nest success. High annual variability in repro- 
duction appears to be characteristic of all goshawk 
populations studied to date and is associated with 
annual variation in weather and prey (Kostrzewa 
and Kostrzewa 1990, Keane 1999, Doyle and Smith 
2001). Finally, research on long-lived raptors sug- 
gests that some breeding areas consistently fledge 
more young than others, with the majority of 
young in the population being produced by a few 
females that occupy high-quality breeding areas 
(e.g., Newton 1989, 1991). Relations between and 
among productivity, habitat quality, population 
size, and trends in goshawks are not clear, and ob- 
served trends in productivity by themselves cannot 
be related to population status. As a result, it is 
difficult or impossible to discern any trends in gos- 
hawk reproductive success in the recent past over 
a wide geographic area. However, even if such tem- 
poral trends were discernable in the western U.S., 
such trends per se would not serve as an adequate 
foundation for concluding that similar trends 
would thereby exist in population size. Informa- 
tion on reproduction must be combined with sur- 
vival and immigration-emigration data at appro- 
priate scales to derive population growth rates 
(e.g., Maguire and Call 1993). To date, such infor- 
mation on goshawks in the western U.S. does not 
exist. 

Distribution. Squires and Reynolds (1997) provid- 
ed the most current delineation of known year- 
round and wintering ranges of goshawks in the 
western U.S. Contraction of historical breeding or 
wintering ranges could suggest a decline in popu- 
lation size (Kennedy 1997), but no historical or 
current evidence is available to suggest either a 
range contraction or expansion in the western U.S. 
Without reliable information on historical breed- 
ing and wintering ranges, knowledge of current 
ranges has limited utility to evaluate current pop- 
ulation size or trends. 

Encounter rates-detection surveys. Surveys for nest- 
ing goshawks in the western U.S. have been con- 
ducted in anticipation of proposed timber sales. 
While some land-management agencies adhere to 
established survey protocols (e.g., Kennedy and 
Stahlecker 1993, Joy et al. 1994), many have not, 
resulting in spatial and temporal variation in meth- 


odology. Techniques that do not detect all gos- 
hawks present have not been validated by estimat- 
ing density at multiple sites with known breeding 
densities (presumably all methods except complete 
searches of survey plots; even with complete 
searches, multiple years are probably necessary to 
detect all goshawk pairs present [DeStefano et al. 
1994a, Reynolds and Joy 1998]). Thus, goshawk de- 
tection rates and estimated nest densities generally 
cannot be directly compared spatially, or even tem- 
porally at the same site. 

Nest density and detection rates from surveys are 
also influenced by how study areas are defined and 
located (Smallwood 1998). The primary purpose 
of most goshawk surveys is not to estimate breed- 
ing densities or population parameters, but to lo- 
cate nests for protection and to predict or mitigate 
the effects of proposed timber sales on goshawks. 
As a result, the locations of surveys for goshawks 
are generally not random with respect to potential 
goshawk habitat. Thus, the results from such sur- 
veys can appropriately be applied to the goshawk 
nests studied, but any inference beyond the sample 
is speculative. Comparing among studies is also in- 
appropriate in some cases because of differences 
in survey techniques, interpretation, and report- 
ing. Inconsistent definition and use of terms relat- 
ed to goshawk ecology (see the Appendix for pro- 
posed standard terminology) further confound 
comparisons among studies. These factors limit the 
utility of detection surveys as an index to goshawk 
densities and population trends in the western U.S. 
Existing data from detection surveys do not pro- 
vide insight into goshawk population status beyond 
documenting occurrence of breeding birds at sur- 
vey sites. 

Demographic data. Demographic studies often fo- 
cus on estimating lambda (the annual rate of pop- 
ulation growth) with Leslie-matrix projection mod- 
els from estimates of age-specific fecundity and 
survival. However, even at the scale of local study 
areas, data necessary to estimate lambda are gen- 
erally inadequate for goshawks (e.g., DeStefano et 
al. 1994b, Reynolds and Joy 1998). While consid- 
erable information exists on fecundity, there are 
few estimates of adult survival, and data on juvenile 
survival are lacking (but see DeStefano et al. 
1994b, Kennedy 1997, Ingraldi 1998, Reynolds and 
Joy 1998). With the possible exception of the on- 
going long-term study on the Kaibab Plateau in Ar- 
izona (Reynolds and Joy 1998), studies have not 
been conducted for sufficient time periods with ad- 


196 


Andersen et al. 


VoL. 39, No. 3 


equate sample sizes to understand temporal varia- 
tion in adult survival and fecundity. The propor- 
tion of adults attempting to breed has been 
estimated in only a few places (Reynolds and Joy 
1998). Among-year movements, especially by adult 
female goshawks to different nesting areas, add 
complexity to estimating demographic parameters, 
because without radiotelemetry data, the fate of 
these birds will often be unknown (Flatten et al. 
2001 ) . Production of young (to fledging) has been 
estimated in a number of studies, but only in a few 
locations have these data been coupled with sur- 
vival information. Finally, information regarding 
immigration and emigration of juvenile and adult 
goshawks is lacking. Thus, while demographic stud- 
ies have significantly increased understanding of 
goshawk population dynamics, no studies to date 
have generated adequate empirical stage-specific 
estimates of survival and fecundity for estimating 
lambda with matrix projection models at local 
scales, and demographic data are unavailable at 
larger scales, making it impractical to estimate 
lambda for the western U.S. Recent alternative 
models for estimating lambda (e.g., Pradel 1996), 
or models for assessing trends in adult survival, 
have not been applied to existing goshawk data, 
but they should be explored. 

Direct estimation of trends in breeding popula- 
tion size on local study areas has been hampered 
by problems associated with searching large areas 
for nests, difficulty in detecting pairs that are pres- 
ent but not nesting, edge effects, limited method- 
ology available to estimate density, and spatial and 
temporal variation in search effort and protocols. 
In addition, size and location of study areas can 
affect estimation of population size (Smallwood 

1998) because study areas are seldom chosen ran- 
domly. Thus, similar to estimating population 
growth rate based on demographic rates, estimat- 
ing population trends on the scale of local study 
areas has had limited success. 

Trends in density. Breeding densities of goshawks 
vary considerably across their geographic range; 
densities in 10 published studies in North America 
ranged from 0.03-11.9 pairs or nests per 100 km^. 
In the western U.S., excluding Alaska, densities in 
seven published studies ranged from 1.4-11.9 pairs 
or nests per 100 km^ (Squires and Reynolds 1997, 
Reynolds and Joy 1998, FWS 1998, Bosakowski 

1999) . Goshawk density (number of breeding 
pairs/area) reported in unpublished work sum- 
marized by the FWS (1998) fell within the same 


range. Comparison among existing estimates of 
breeding density are confounded by a number of 
factors, including variation among studies in defi- 
nitions of densities, territories, pairs, “active” 
nests, and occupied nest areas (see Appendix) . In 
addition, the small number of published studies of 
goshawk breeding density {N — 7) , the limited du- 
ration of most studies (median = 2.0 yr; Squires 
and Reynolds 1997), and high temporal variability 
in reproduction preclude reliable assessment of 
temporal trends in breeding densities of goshawks 
across the western U.S. The logistical problems of 
determining density in goshawks and possible 
methodological bias in selecting nest search areas 
for some studies (Kennedy 1997, Squires and Reyn- 
olds 1997, Smallwood 1998, Trexel et al. 1999) may 
further confound analyses of breeding densities as 
an index to population size. Moreover, densities of 
the nonbreeding segment of goshawk populations 
(floaters) and their demographic role are entirely 
unknown (Hunt 1998) . Theoretically, a population 
decline may occur without concurrent decline in 
nesting densities if floaters are available to fill va- 
cant breeding territories. Declines in nesting den- 
sity may only then become apparent after the float- 
er population has been exhausted (Franklin 1992). 
Currently, existing data on nesting and breeding 
densities are not adequate to assess goshawk pop- 
ulation trends across western North America. 

Historical records. There have been no systematic 
efforts to synthesize existing historical goshawk rec- 
ords across North America, and only limited infor- 
mation is available for portions of their range (e.g., 
Grinnell and Miller 1944); therefore, historical 
data were not available to the FWS for assessing 
change in goshawk distribution in the western U.S. 
Use of historical records for assessing distributional 
change has limitations because natural history col- 
lections are not random or systematic samples 
from across the historical range of a species (Shaf- 
fer et al. 1998). The number of historical goshawk 
records represented in museum collections is also 
limited because of the relative rarity of goshawks, 
their secretive behavior, and predominant occur- 
rence in remote locales. Because of these limita- 
tions, historical records are not available for as- 
sessing historical ranges and current changes in 
distribution for goshawks in all regions of North 
America. Data necessary to assess historical gos- 
hawk distribution across western North America 
have not been collected, and thus contrasts be- 


September 2005 


Status 


197 


tween historical and current ranges of goshawks in 
the western U.S. are only possible for limited areas. 

Genetic Structure. Observed morphological patterns. 
Two subspecies of goshawks {A. g. atricapillus, A. g. 
laingi) were recognized in the western U.S. and 
southeast Alaska by the American Ornithologists’ 
Union in 1957 (AOU 1957). A. g. atricapillus occnv^ 
across nearly all forested regions of the western 
U.S., Canada, the western Great Lakes region, and 
the northeastern U.S. A. g. laingi occurs from Van- 
couver Island, insular British Columbia, to the Al- 
exander Archipelago of southeastern Alaska 
(Whaley and White 1994). A third, putative sub- 
species (e.g., Stresemann and Amadon 1979), A. g. 
apache, occurs in the mountains of southern Ari- 
zona, but was not recognized by the AOU (1957) 
and is currently not recognized by most taxono- 
mists (Whaley and White 1994) . Morphological dif- 
ferences between eastern and western A. g. atricap- 
illus have not been demonstrated in the literature 
(see Whaley and White 1994). Ridgway (in Baird 
et al. 1875) speculatively divided eastern {Astur atri- 
capillus atricapillus) and western (then termed Astur 
atricapillus striatulus) goshawks, but others, includ- 
ing Taverner (1940), have not made this distinc- 
tion. Sample sizes have been small in the analyses 
of eastern A. g. atricapillus, or the analyses were 
confounded by migrants (Mueller et al. 1976). 
Since Whaley and White (1994), there have not 
been any in-depth analyses of A. g. atricapillus 
across the continent using larger sample sizes. 

Genetic population structure. There are few publi- 
cations on the phylogeography of DNA in North 
American goshawks. In an unpublished report, 
Gavin and May (1996) did not detect genetic dif- 
ferences among goshawks representing A. g. atri- 
capillus, A. g. laingi, and A. g. apache. More recently, 
Sonsthagen et al. (2004) used eight microsatellite 
DNA loci and mitochondrial DNA control-region 
sequence data to assess population structure of 
goshawks breeding in Utah. Their pairwise com- 
parisons using microsatellite markers found no dif- 
ferentiation among the sampled sites {N = 49 
birds) from northern to southern Utah. Overall, 
they found low levels of population structuring. 

During the 1990s, numerous goshawk tissue sam- 
ples were collected from Arizona to Alaska, and 
many of these samples have been analyzed to eval- 
uate genetic variation in North American gos- 
hawks. Preliminary data from markers assayed 
from goshawks nesting in Alaska (coastal and in- 
terior), British Columbia (coastal and interior). 


and Utah suggest that genetic differences in pop- 
ulations will be found as analyses are completed. 

Western goshawks as a discrete population. In the 
context of the Endangered Species Act, a discrete 
population of a vertebrate species is one that sat- 
isfies at least one of the following conditions: (1) 
it is markedly separated from other populations of 
the same taxon as a consequence of physical, phys- 
iological, ecological, or behavioral factors or (2) it 
is delimited by international boundaries within 
which differences in control of exploitation, man- 
agement of habitat, conservation status, or regu- 
latory mechanisms exist that are significant in light 
of section 4(a) (1) (D) of the Act (USDI and United 
States Department of Commerce 1996). Goshawks 
that breed in the western and eastern U.S. are part 
of a continuous population that extends across 
Canada and into interior Alaska but that is seg- 
mented by international boundaries (Squires and 
Reynolds 1997). It was beyond the scope of our 
charge to assess differences in management of gos- 
hawks in the U.S. and Canada, and there is cur- 
rently little evidence of biological differences be- 
tween goshawks in the eastern and western U.S. 
Therefore, it is unclear whether goshawks breed- 
ing in the western and eastern U.S. should be 
viewed as discrete population segments under fed- 
eral threatened and endangered species policy. 

Habitat Relations. Long-term forest-management 
patterns. It is likely that past and current forest 
management on public and private lands in the 
western U.S. has resulted in existing landscapes 
that are quite different from historical landscapes 
and their natural range of variation. It was beyond 
the scope of our charge to project the condition 
and attributes of future forested landscapes in the 
western U.S. Clearly, though, forested landscapes 
that contain goshawk habitat will be necessary to 
support goshawk populations in the future. In 
1998, the FWS (USDI 1998) concluded that cur- 
rent and projected land-management practices in 
the review area would not result in landscapes in- 
capable of supporting goshawks. This conclusion 
was predicated on both an assessment of future 
landscape condition and goshawk response to that 
condition, both of which were speculative. 

Status of prey populations. Across western North 
America, goshawks feed on a variety of prey spe- 
cies, including birds and mammals from small to 
moderately large in size. Passerines (primarily 
corvids and thrushes [Turdidae]), woodpeckers 
(Picidae), Galliformes (grouse, ptarmigan, quail). 


198 


Andersen et al. 


VoL. 39, No. 3 


tree (Sciurus spp.) and ground squirrels (Spermo- 
philus spp.), and lagomorphs (including snowshoe 
hares [Lepus americanus] and cottontail rabbits [5))Z- 
vtlagus spp.]) are the major prey species. Almost 
all information (but see Beier and Drennan 1997, 
Drennan and Beier 2003) regarding prey use of 
goshawks is derived from studies of nests during 
the breeding season, and it is based on observa- 
tions of prey delivered to nests, prey remains col- 
lected at nests, or pellets and remains collected at 
nests or plucking perches. These data may primar- 
ily reflect prey selection by male goshawks, which 
provide most of the food during pre-incubation 
through fledging. Further, most studies report on 
the frequency of prey species pooled across years. 
Only a few North American studies have assessed 
annual variation in diet and related it to variation 
in demographic parameters, such as reproduction 
(e.g., Keane 1999, Maurer 2000, Doyle and Smith 
2001). Diets during winter may differ from diets 
during the breeding season (e.g.. Widen 1989) be- 
cause of prey hibernation, goshawk migration, or 
changes in use of vegetation types by prey species 
or goshawks in different bioregions. Little infor- 
mation exists on winter diets for goshawks in west- 
ern North America (Squires and Reynolds 1997). 

In the western U.S., most diet studies report that 
prey associated with late-successional forests are 
important (Reynolds and Meslow 1984, Kennedy 
1991, Reynolds et al. 1992, Keane 1999, Maurer 
2000, Lewis 2001), although species associated with 
other forest age classes or vegetation associations 
are also used (e.g., Reynolds et al. 1992, Boal and 
Mannan 1994, Doyle and Smith 1994, Younk and 
Bechard 1994, Patla 1997, Watson et al. 1998). Al- 
though a large number of species are usually re- 
corded in overall summaries of prey species, par- 
ticular species or a smaller suite of prey species 
make a relatively greater contribution to total bio- 
mass and have been associated with temporal var- 
iation in reproduction. Further, these important 
prey species, or suites of prey species, vary among 
bioregions or major vegetation types (Reynolds et 
al. 1992, Watson et al. 1998, Keane 1999, Doyle and 
Smith 2001). 

Although considerable information exists about 
diet of goshawks during the breeding season, the 
relations between goshawks and prey abundance, 
availability, and distribution in the landscape are 
difficult to study and will not be well understood 
in the near future, at least at the scale of the west- 
ern U.S. Considerable additional information re- 


garding the impacts of future forest conditions in 
the western U.S. on goshawk prey species is re- 
quired before goshawk population responses to 
trends in prey abundance resulting from forest- 
management practices can be assessed. 

Association of goshawks with habitat at multiple spa- 
tial scales. Goshawk-habitat relations have been in- 
vestigated at a number of spatial and temporal 
scales. There is general agreement among biolo- 
gists that goshawk breeding habitat can be dis- 
cussed in terms of three nested spatial scales: a nest 
stand (and alternative nest stands; 10-12 ha), a 
post-fledging area (PFA; 120-240 ha), and a for- 
aging area (1500-2100 ha; Reynolds et al. 1992). 
Considerable information exists regarding charac- 
teristics of nest trees, but comparatively fewer data 
exist on habitat use outside of the breeding season. 

Breeding Season. Nest tree. Goshawks build and 
use nests in a variety of conifer and hardwood tree 
species. They often use trees that are among the 
larger or largest in the stand (e.g., Keane 1999). 
Common nest-tree species include ponderosa pine 
{Pinus ponderosa) in the southwestern U.S,, Doug- 
las-fir (Pseudotsuga menziesii) and other conifers in 
the Rocky Mountains, Sierra Nevada, Pacific 
Northwest, and Alaska, and aspens (Populus spp.) 
in portions of the Rockies and interior Alaska. 
Squires and Reynolds (1997:6) concluded that gos- 
hawks “tend to nest in a relatively narrow range of 
vegetation structural conditions,” suggesting that 
tree species used for nesting is secondary to struc- 
tural characteristics of the tree and surrounding 
vegetation. 

Nest stand. A nest stand is that area covered by a 
forested patch consisting of trees often character- 
ized by similar size, species, and spacing, in which 
a goshawk nest is located. Studies of nests and nest 
stands have been widespread, covering much of 
the goshawk’s range in the western U.S. Stands 
where tree species such as ponderosa pine or 
lodgepole pine {P. contorta) predominate and 
stands of mixed conifer species are used for nest- 
ing. Aspen stands in mountain valleys and draws in 
the Great Basin of Nevada and Oregon are also 
used for nesting. Most studies of goshawk nest 
stands have focused on forest structure (e.g., Reyn- 
olds et al. 1982, Moore and Henny 1983, Hayward 
and Escano 1989, Daw et al. 1998) in the vicinity 
of the nest tree and indicate that large trees and 
well-developed canopies are important. The spe- 
cies of tree used for nesting or those that constitute 
the nest stand appear to be less critical. Goshawks 


September 2005 


Status 


199 


usually nest in stands of late-successional forest, 
where trees are often larger than those of other 
forested stands nearby (e.g., Reynolds et al. 1982). 
Habitat composition within these nesting stands 
may include single canopy or multi-story layer 
components. Forest management that fragments 
and reduces the extent and area of stands suitable 
for nesting in a breeding area may result in its less 
consistent use for nesting over time (e.g., Wood- 
bridge and Detrich 1994, Desimone 1997). 

Across the western U.S. and Alaska, many studies 
have documented goshawks selecting nest stands 
that are more mature or consist of late-successional 
forest compared with random assessments of near- 
by forest habitat, irrespective of scale of analysis 
(e.g., Moore and Henny 1983, Crocker-Bedford 
and Chaney 1988, Desimone 1997, Keane 1999). 
Some studies have suggested that high-canopy clo- 
sure is one of the more uniform characteristics of 
goshawk nest stands (Hayward and Escano 1989, 
Keane 1999), and others have documented that a 
higher percent canopy closure was associated with 
a higher probability that goshawks would nest in a 
stand (Crocker-Bedford and Chaney 1988). Cano- 
py closure in nest stands is variable across North 
America, and in some regions of the western U.S. 
and Alaska mean canopy closure near the nest 
might be rather low (ca. 50% in parts of Oregon 
and Washington [McGrath 1997] and southeastern 
Alaska [Iverson et al. 1996]). Differences in sam- 
pling methods probably account for some of this 
apparent inconsistency because measurement of 
canopy closure has not been conducted consis- 
tently among studies (Crocker-Bedford and Cha- 
ney 1988). However, even where canopy closure 
around a nest area is apparently low, it is still gen- 
erally higher than the surrounding portions of the 
stand or other nearby stands. This suggests that 
high-canopy closure relative to the range of avail- 
able canopy closure might be more important than 
absolute canopy closure, at least above some min- 
imum threshold. 

Why goshawks select stands with relatively larger 
trees and higher canopy cover is not known. Po- 
tential hypotheses include; (1) increased protec- 
tion from predators, (2) increased food availability, 
(3) reduced exposure to cold temperatures and 
precipitation during the energetically stressful pre- 
laying period in late winter-early spring, (4) re- 
duced exposure to high temperatures during the 
summer nestling period, (5) reduced competition 
with raptor species that nest in more open envi- 


ronments (e.g., Red-tailed Hawk {Buteo jamaicen- 
^A]), or (6) increased mobility because of reduced 
understory vegetation in mature stands. 

Use area-home range. How goshawks use habitats 
away from their nests during the nesting season is 
not well understood. Methods to evaluate gos- 
hawk-habitat associations at the home-range scale 
fall into a few different categories, including: (1) 
habitat evaluations based on circular areas cen- 
tered on the nest that are often made using aerial 
photography or other remote sensing methods and 
Geographic Information Systems, (2) habitat-selec- 
tion studies using radiotelemetry, (3) evaluating 
hunting habitat use with radiotelemetry and direct 
observation, and (4) evaluating patterns associated 
with habitat disturbance and logging versus fre- 
quency of nesting. 

Most studies of habitat use based on a nest-cen- 
tered evaluation have loosely linked the scale of 
measurement to a nest stand, PFA, or mean home- 
range size. In general, the preponderance of late- 
successional forest in the landscape decreases as 
the scale increases (i.e., as one moves from nest 
stand to PFA to foraging area; Iverson et al. 1996, 
Finn 2000, Daw and DeStefano 2001, Finn et al. 
2002, McGrath et al. 2003). 

Radiotelemetry studies to evaluate habitat use 
within the home range during the nesting season 
have found that goshawks selected for late-succes- 
sional forests even beyond their nest stands (Widen 
1989, Austin 1993, Bright-Smith and Mannan 1994, 
Hargis et al. 1994, Iverson et al. 1996, Beier and 
Drennan 1997). Goshawks used larger stands of 
late-successional forest than was available in south- 
eastern Alaska (Iverson et al. 1996, Pendleton et 
al. 1998) and Sweden (Widen 1989); in Arizona, 
some goshawks selected for late-successional forest 
>200 m from openings (Bright-Smith and Mannan 
1994) . In California, goshawk locations had greater 
basal area, canopy cover, and more large trees than 
did random points (Austin 1993, Hargis et al. 
1994). These results suggest a fine-scale selection 
for larger stands of mature forests within goshawk 
nesting-season home ranges. 

Presumably, vegetative characteristics associated 
with foraging sites influence prey availability. For 
example, Beier and Drennan (1997) concluded 
that goshawks in Arizona did not select foraging 
sites based on prey abundance; rather, they select- 
ed sites based on vegetation structure. Goshawk 
foraging locations had a higher canopy closure, 
greater tree density, more large trees, and fewer 


200 


Andersen ex al. 


VoL. 39, No. 3 


shrubs and saplings than random reference plots. 
There was also selection for dense stands with high 
canopy closure that were rare on their study area. 
Widen (1989) had previously reported that in Eu- 
rope, hunting sites were associated with habitat 
structure and did not seem to be related to abso- 
lute prey abundance. A number of authors have 
noted that foraging sites typically are characterized 
by open space between the bottom of the canopy 
and the top of the shrub layer (e.g., Reynolds 1989, 
Widen 1989, Crocker-Bedford 1990, 1998, Beier 
and Drennan 1997) and have speculated that this 
space may increase prey vulnerability by providing 
a flight path for foraging goshawks. 

Results of several studies suggest that goshawks 
are more likely to reoccupy breeding areas within 
landscapes that have larger proportions of late-suc- 
cessional forest, compared with landscapes that 
have smaller proportions of these forests (Ward et 
al. 1992, Woodbridge and Detrich 1994, Daw 1997, 
Patla 1997, Finn 2000, Finn et al. 2002). Widen 
(1997) concluded that intensive forest manage- 
ment was the prime factor in reductions in gos- 
hawk breeding density across nine study areas in 
Scandinavian boreal forests. 

Assessing habitat use at the home-range-use 
area scale has several important limitations, includ- 
ing small sample sizes, variation in fecundity, and 
the small range of vegetation types in which these 
studies have been conducted. In addition, consid- 
erable variation likely exists among home range- 
use areas, with some use areas consistently produc- 
ing young, and others only occasionally producing 
young (Newton 1989, Joy 2002, McClaren et al. 
2002). Thus, habitat evaluations that are not relat- 
ed to productivity and population dynamics might 
have limited utility. Including use areas that rarely 
produce young in these evaluations might make it 
difficult to identify characteristics of use areas as- 
sociated with high-quality habitat. Finally, habitat 
use at the home-range scale has been assessed in 
only a few vegetation types, limiting inference to 
scales below that of the western U.S. Clearly, ad- 
ditional information is necessary to better assess 
habitat use patterns at the scale of home range- 
use areas. 

Non-nesting season. There are few studies of gos- 
hawk-habitat associations during the non-nesting 
season in North America. Iverson et al. (1996) ex- 
amined year-round habitat selection by radio- 
tagged adult goshawks in southeastern Alaska with- 
in their seasonal use area and found no differences 


in habitat selection between the nesting season 
and non-nesting season. Adult goshawks selected 
for larger size classes of late-successional conifer- 
ous forest compared with other habitat cover types. 
Beier (1997) and Drennan and Beier (2003) ex- 
amined winter foraging habitat of adult goshawks 
in northern Arizona and found that goshawk lo- 
cations were in areas with a slightly higher medi- 
um-size tree density and higher canopy cover than 
contrast plots. Females remained in the ponderosa 
pine vegetation type, and most males moved to 
pinyon-juniper (Pinus-Juniperus) woodlands. Some 
goshawks move to open or scrub habitats in the 
winter (Squires and Ruggiero 1995), while others 
seem to remain in forested areas, making it diffi- 
cult to generalize across populations in terms of 
goshawk winter-habitat use. 

Summary of goshawk habitat use. Goshawks have 
broad geographic and elevational distributions in 
North America and can be found in many different 
forest types and forest stand conditions (Squires 
and Reynolds 1997). Goshawks have relatively large 
home ranges, are able to move great distances— es- 
pecially dnring times of low prey abundance, and 
use a wide variety of prey species across the range 
of landscapes in which they occur. Goshawks tend 
to nest in forest stands with specific structural char- 
acteristics, such as stands with large trees and mod- 
erate to high canopy closure that is high relative 
to the range of available canopy closure. Goshawks 
forage in a variety of habitats, ranging from early- 
successional forests, to mature forests, to open hab- 
itats adjacent to forested habitats. During the 
breeding season, late-successional forests appear to 
be used predominantly for foraging, although 
some of the prey taken by goshawks use young for- 
ests and open habitats. 

Goshawk breeding habitat can be discussed in 
terms of three nested, spatial scales: a nest stand 
(and stands containing alternative nests), within a 
PFA, and within a foraging area. At the nest-stand 
scale, late-successional forest characteristics are of- 
ten important determinants of where goshawks lo- 
cate their nests. The preponderance of late-succes- 
sional forest in the landscape decreases as the scale 
increases (e.g., as one moves from nest stand to 
PFA to foraging area), and existing data from te- 
lemetry and observational studies suggest that gos- 
hawks use late-successional forests within their 
home ranges for foraging, but use prey associated 
with both early- and late-successional forests, and 
in some cases, open habitats. Thus, goshawks ap- 


September 2005 


Status 


201 


pear to be associated with late-successional forests 
for nesting and foraging, but clearly also use, and 
use prey associated with, other cover types. Gos- 
hawk breeding habitat has been studied much 
more intensively than nonbreeding habitat. In 
some landscapes, goshawks appear to remain near 
breeding areas throughout the year, although 
there is considerable annual variation and varia- 
tion between sexes in nonbreeding habitat use. In 
at least some landscapes, goshawks forage in late- 
successional forest habitats throughout the year. 
Conversely, some goshawks use landscapes during 
the nonbreeding season (e.g., pinyon^uniper and 
open sagebrush basins) that are quite different 
from landscapes used during the breeding season. 
In general, there appears to be a wider range of 
habitats used during the non-breeding season than 
during the breeding season. 

Habitat as a Surrogate for Population Trends. 
Context. The population status of goshawks and 
their association with late-successional forests in 
western North America has been debated for >10 
yr. This debate has considerable bearing on the 
FWS decision that listing goshawks in the western 
U.S. under the Endangered Species Act was not 
warranted (USDI 1998). In 1990, Crocker-Bedford 
(1990) reported a relation between timber harvest 
and loss of goshawk territories on the Kaibab Pla- 
teau in Arizona and suggested that some forest- 
management practices might negatively affect gos- 
hawk populations. Considerable discussion of that 
conclusion and the evidence supporting it ensued. 
Kennedy (1997) later reviewed the status of gos- 
hawks and concluded that data were lacking to de- 
termine if populations of goshawks were increas- 
ing, decreasing, or stationary. She called for more 
in-depth demographic studies, including meta- 
analysis approaches, combining ongoing studies 
with marked goshawks. Smallwood (1998) and 
Crocker-Bedford (1998) both responded to Ken- 
nedy’s review paper. Smallwood (1998) suggested 
that in lieu of appropriate sampling and agree- 
ment among scientists regarding additional vari- 
ables that should be analyzed, evidence for a gos- 
hawk population decline should be based on 
availability and contiguity of habitat and migration 
counts. Crocker-Bedford (1998) hypothesized that 
distribution of foraging habitat across the land- 
scape influenced goshawk home-range size, which 
in turn influenced breeding pair density and re- 
productive success. He suggested further devel- 
opment of models of goshawk-habitat relations, 


inventory of current forest conditions, and assess- 
ment of population status based on habitat condi- 
tions at the landscape level. 

In their status review (FWS 1998), the FWS at- 
tempted to assess population status from popula- 
tion data and also by using the distribution and 
extent of habitat, particularly older forest (specifi- 
cally old-growth), as a surrogate for a direct mea- 
sure of population trends. This effort represented 
the largest concerted attempt to date to document 
goshawk locations and habitat in North America. 
The FWS concluded that it was evident that “there 
[are] inadequate data available which could be 
used to determine the population trend for gos- 
hawks throughout the review area. Furthermore, 
our knowledge of the factors that affect the size of 
goshawk populations at local and regional levels, 
or in the entire area is incomplete. A clearer un- 
derstanding of population size and factors affect- 
ing goshawk populations is needed. Much of what 
is known is currently applicable only to local pop- 
ulations and localized habitat conditions and ef- 
fects, and should not be extrapolated to the larger 
range of the species” (FWS 1998). The FWS also 
noted that few studies have focused on goshawk 
population dynamics over a sufficient period of 
time to provide the kinds of demographic data 
needed for a status review. With this realization, 
FWS attempted to identify trends in habitat. The 
FWS concluded that they could not directly tie 
changes in goshawk populations to changes in hab- 
itat over time because of a lack of data and little 
confidence regarding how goshawk populations re- 
spond to changes in their habitat. The FWS deci- 
sion that listing goshawks in the western U.S. un- 
der the Endangered Species Act was not warranted 
was based in large part on lack of evidence that 
habitat was currently limiting goshawks, and that 
habitat was unlikely to limit the goshawk popula- 
tion in the review area in the foreseeable future. 
Such an approach is clearly limited by how well the 
relations between goshawks and their habitat are 
understood, and how well existing vegetative con- 
ditions are known. 

Existing goshawk-habitat models. Warren et al. 
(1990) developed a goshawk-habitat model based 
on a review of published and unpublished litera- 
ture and expert opinion using the Delphi method. 
In their model, habitat suitability increased with 
increasing canopy cover, size of overstory trees, size 
of the nest stand, and decreasing slope. Suitability 
of foraging habitat was modeled in relation to prey 


202 


Andersen et al. 


VoL. 39, No. 3 


availability, forest type, and tree species composi- 
tion. Reynolds et al. (1992) synthesized habitat as- 
sociations for goshawks and 14 prey species and 
silvicultural prescriptions designed to produce suit- 
able forest conditions for goshawks and their prin- 
cipal prey in the southwestern U.S. Such prescrip- 
tions were developed with the intent of (1) 
sustaining goshawk populations in the Southwest, 
(2) providing desired forest conditions for the gos- 
hawk and its prey, (3) using the natural, presettle- 
ment forest composition, structure, and landscape 
pattern of each forest type as a template for assem- 
bling, and assuring the sustainability of, goshawk 
and prey habitats in large landscapes, and (4) man- 
aging southwestern forests as an ecosystem (i.e., re- 
taining all of the parts) . For the goshawk, this is a 
conceptual model, but the recommendations that 
came from this model are being implemented on 
national forests throughout the Southwest while 
components of the model are being implemented 
throughout much of the western U.S. and in Brit- 
ish Columbia, Canada. The model of Reynolds et 
al. (1992) has served as the primary model for gos- 
hawk management in the southwestern U.S. (Reyn- 
olds et al. 1996, Long and Smith 2000) and has 
been the subject of considerable debate and eval- 
uation (e.g., Braun et al. 1996). 

In Utah, Johansson et al. (1994) used elevation 
and vegetation models to predict potential gos- 
hawk nesting sites. They found elevation to be a 
better predictor of goshawk nest locations than 
vegetation, although elevation, vegetation, and veg- 
etative characteristics of PFAs were the best predic- 
tors overall. In Idaho, Lilieholm et al. (1994) ap- 
plied a stand density index (SDI) — a measure of 
stand density that is based on mean tree size and 
density and is comparable among stands — to guide 
management practices intended to create forest 
conditions similar to those found in goshawk nest 
areas. Although this latter method was primarily 
intended to assist silviculturalists in managing for- 
est stands, mean tree size and density of stands rep- 
resenting goshawk habitat (e.g., goshawk nest 
areas) can be used as models of desired future con- 
ditions. Similarly, Graham et al. (1994) pointed out 
that the way forests regenerate, develop, and die is 
highly variable in time and space, and recom- 
mended managing large tracts of forests as sustain- 
able ecological units rather than managing smaller 
tracts as individual home ranges. DeStefano (1998) 
suggested that goshawk occurrence was related to 
characteristics associated with late-successional for- 


est, but that goshawks are found in a wide variety 
of forest conditions. Crocker-Bedford (1998) hy- 
pothesized that distribution of foraging habitat 
across the landscape influences goshawk home- 
range size, which in turn influences breeding pair 
density and reproductive success. Landscapes that 
contain a higher concentration of foraging habitat 
with adequate prey abundance should support 
higher densities of breeding goshawks. 

Joy (2002) developed spatial-simulation models 
to assess the relations between goshawk habitat 
composition and structure and the location of 
nests and use areas and the relations between the 
amount and arrangement of habitat components 
in high- and low-quality breeding areas. High- and 
low-quality breeding areas were distinguished 
based on long-term (10 yr) demographic data from 
101 breeding areas in northern Arizona. Joy 
(2002) found that intraspecific territoriality plays a 
more significant role in nest location than avail- 
ability of nest area habitat on the Kaibab Plateau. 
In addition to using habitat models to identify spa- 
tial and compositional differences between gos- 
hawk nests and random locations, Joy (2002) and 
Reich et al. (2004) used these models to predict 
nest locations likely to have high reproductive out- 
put. 

McGrath et al. (2003) developed models relating 
habitat characteristics around goshawk nest sites at 
scales from 1-170 ha in eastern Oregon and Wash- 
ington. At the 1-ha scale, structural stage (i.e., late- 
seral), topographic position (i.e., lower slopes and 
drainage bottoms), and stand-basal area (i.e., high 
basal area) were associated with goshawk nests, 
with high basal area being the most important. At 
larger scales (10-170 ha), later serai stages, high 
understory growth, and high canopy closure were 
associated with nests and these associations were 
prevalent up to 83 ha. They concluded that: (1) 
there is a core area around goshawk nests where 
the forest is generally mid- to late-successional 
stage (large trees with high canopy closure) and 
(2) this core is surrounded by diverse types of for- 
est cover that are equally abundant (i.e., no one 
cover type dominates). 

In summary, most existing models of goshawk- 
habitat relations are limited to vegetative structure 
used for nesting. Other habitat variables (such as 
microclimatic conditions at nest, foraging, or roost 
sites) and other aspects of life history (such as ju- 
venile dispersal and territory establishment, non- 
breeding or failed breeding adults, and winter 


September 2005 


Status 


203 


ecology) have received relatively little attention 
compared to vegetative structure around nests, 
largely because of the difficulties in working with 
goshawks in the field. 

Limitations on using current goshawk— habitat models 
for predicting population status. Currently, the rela- 
tions between goshawks and their habitat in the 
western U.S, are not understood well enough to 
use trends in habitat as a surrogate for trends in 
goshawk populations. Fundamentally, this is be- 
cause there are unknown functional relations 
among the amounts and distribution of goshawk 
habitat, the range of vegetation conditions that 
characterize goshawk habitat, and goshawk popu- 
lation densities and population dynamics. There- 
fore, it is not currently possible to predict how 
changes in habitat, or changes in specific types of 
vegetation such as old-growth forests, are related 
to changes in goshawk population densities or 
trends. The use of late-successional forests (specif- 
ically old-growth forest) as a surrogate for goshawk 
population status is limited because: (1) goshawks 
show a high degree of versatility in habitat use, and 
although late-successional forest is a commonly 
used habitat, other serai stages also are used; thus, 
reliance on distribution of late-successional forests 
alone for determining the status and distribution 
of goshawks in the western U.S. is not sufficient; 
(2) important prey species vary among bioregions 
and major vegetation types with late-successional 
forest associates (e.g., Douglas [Tamiasciurus doug- 
lasii\ and red squirrels [T hudsonicus^) important 
in some regions and early-seral species (e.g., snow- 
shoe hares) relatively more important in other re- 
gions; (3) there is currently no consistent defini- 
tion of old-growth forest as it pertains to goshawk 
habitat that can be applied across the entire west- 
ern U.S. or at the scale of major vegetation types; 
(4) habitat may not be occupied if factors other 
than old-growth vegetative structure (e.g., weather, 
prey availability) are limiting goshawk populations; 
and (5) large-scale, regional vegetation mapping 
efforts (e.g., msyor portions of the western U.S.) 
are not sufficiently precise or accurate to assess 
current or future conditions. Multiple factors influ- 
ence habitat use, especially on very large spatial or 
temporal scales, and relations between goshawks 
and habitats and goshawks and prey species, seem 
to be variable across vegetation types. Knowledge 
concerning the functional relation between the 
distribution and abundance of habitat and gos- 
hawk population densities and trends is required 


in order to draw scientifically defensible inferences 
regarding how changes in habitat, or specific hab- 
itat types such as old-growth, relate to changes in 
goshawk populations. Currently this relation is un- 
known. 

Recommendations 

To assess goshawk population status in the west- 
ern U.S. or any other portions of this bird’s range 
in North America, several improvements in exist- 
ing data-collection efforts and protocols are nec- 
essary. Additional data that do not currently exist 
will also need to be collected before adequate pop- 
ulation status assessment can take place in the west- 
ern U.S. Items we identified include: 

(1) Compilation and accessibility of existing data. We 
urge organization of existing data into a for- 
mat that would make it readily accessible to 
management agencies and other interested 
parties. Development of standardized proto- 
cols for future monitoring and inventory data 
collection will benefit from an assessment of 
the existing information. In addition, devel- 
opment of procedures to systematically and 
regularly capture new information to maintain 
a current database is necessary. 

(2) Sampling strategy. Outside of intensive research 
studies, most existing goshawk distributional 
or occurrence records are based on ad hoc 
sampling generally associated with manage- 
ment activities. If goshawk population trends 
are to be assessed, sampling must represent 
the target population and yield defensible 
trend estimates. Monitoring approaches 
should be based on sample designs that ad- 
dress the definition of the target population, 
appropriate response variable, definition of a 
sampling frame and primary sample units, is- 
sues of probability of detection, and estimates 
of necessary sample sizes required to detect a 
specific change. Monitoring strategies should 
also be designed to assess both population 
trend and habitats, as defined through devel- 
opment of empirical goshawk— habitat rela- 
tions models. Land managers and agency de- 
cision-makers should recognize that continued 
funding of uncoordinated, small-scale goshawk 
monitoring efforts will not yield useful results 
across a large land area. In addition, it may be 
fruitful to address population status at a scale 
smaller than that of the review area. Rather 


204 


Andersen et al. 


VoL. 39, No. 3 


than evaluating goshawk population status for 
the entire western U.S., consideration should 
be given to monitoring trends in goshawk pop- 
ulations and habitat at the ecoregion or biome 
scale (e.g., Sierra Nevada forests; coastal tem- 
perate forests and rainforests of Oregon, 
Washington, and southern coastal British Co- 
lumbia; ponderosa pine forests of New Mexi- 
co, Arizona, and southern Colorado). 

(3) Relation of populations and subspecies. We rec- 
ommend that variation in DNA be used to as- 
sess the phylogenetic relations among eastern 
and western North American A. g. atricapillus, 
and atricapillus to A. g. laingi, and to the pu- 
tative A. g. apache. 

(4) Addressing current limitations of existing data 
sources. Potentially useful data are currently 
limited by a lack of knowledge about popula- 
tion affiliation (e.g., migration counts), small 
sample sizes (e.g.. Breeding Bird Survey data), 
or inadequate sampling strategies. Consider- 
ation should be given to addressing these lim- 
itations where possible. For example, in the 
case of migration counts, population affilia- 
tion of goshawks counted at migration sites 
needs to be determined, perhaps through con- 
servation genetic and stable isotope analyses 
(e.g., Meehan et al. 2001). 

(5) Standardization of terminology^ and protocols asso- 
ciated with estimating breeding status and produc- 
tivity. We recommend that researchers and 
land managers cooperate in developing stan- 
dardized protocols (including terminology 
and data-collection methods) based on peer- 
reviewed literature with the specific intention 
of performing pooled data analysis across the 
entire review area at a later date (e.g., Ander- 
son et al. 1999). If a single set of protocols 
cannot be used for the entire western U.S., 
then standardized protocols should be used 
for large areas (e.g., biomes or ecological hab- 
itat types, but not political boundaries) . 

(6) Research priorities. To assess demography and 
population trends adequately, goshawk-habitat 
relations and the effects of specific land-man- 
agement practices on goshawks in the western 
U.S., coordinated studies of habitat use (possi- 
bly using radiotelemetry) are necessary. Studies 
of demography and habitat use also need to 
address the nonbreeding season, when factors 
regulating populations may be important. In 
addition, land managers need to continue to 


work on remote-sensing applications so that 
broad-scale analysis of habitats such as late-suc- 
cessional forest and patch size can be evaluated. 
Finally, long-term experimental or quasi-exper- 
imental studies are necessary at the landscape 
scale to understand how forest management in- 
fluences goshawks. These studies will be most 
beneficial when accomplished using an inter- 
disciplinary approach in close collaboration 
with land managers. An integrated approach 
between research and management consisting 
of extensive population and habitat monitoring 
at the bioregional scale coupled with intensive, 
long-term demography studies in each of the 
major vegetation types will provide the data 
necessary to monitor goshawk populations and 
habitat and to generate a scientific understand- 
ing of goshawk ecology needed to improve 
management and conservation efforts. 

Finally, we emphasize that in addition to assess- 
ing population trends and status in the western 
U.S., it is also important to better understand gos- 
hawk-habitat relations and the influence of various 
human activities, especially forest-management 
practices, on goshawks. Much of the controversy 
regarding goshawk conservation in the western 
U.S. and elsewhere has to do with concerns about 
forest management and how forest management 
affects goshawks. Thus, it is not sufficient to simply 
assess goshawk population trends in the western 
U.S.; it is also necessary to better understand the 
relations between goshawks and their habitat and 
how human activities affect that habitat. Consid- 
erable information regarding population ecology 
and goshawk-habitat relations currently exists, but 
additional information is necessary. Individual gos- 
hawks or goshawk pairs exhibit landscape-level use 
of space, and thus, occur naturally at relatively low 
densities. They are highly mobile, and as such, 
have proved difficult to study. Thus, a long-term 
investment of resources in a coordinated effort di- 
rected at large spatial scales will be required to as- 
sess goshawk population trends adequately and un- 
derstand goshawk-habitat relations in the western 
U.S. and elsewhere. 

Acknowi.egments 

We express our thanks to Patricia L. Kennedy (Oregon 
State University) and Richard T. Reynolds (U.S. Forest 
Service) for critically reading a prior draft of our report 
and for offering substantive suggestions for improve- 
ment. Dan Edge (TWS), Winni Kessler (TWS), Len Car- 


September 2005 


Status 


205 


penter (TWS), Jim Bednarz (RRF), and Brian Millsap 
(RRF) patiently worked with the committee to meet its 
charge. We also thank Board members of RRF, especially 
Ted Swem, and Council members of TWS, especially Mac 
Baughman, who read this report and offered constructive 
suggestions. Krista E.M, Galley provided copy editing of 
the final report published by TWS. Clayton M. White and 
R. William Mannan provided constructive reviews, and 
Patricia L. Kennedy served as an insightful editor for this 
manuscript. 

Literature Cited 

American Ornithologists’ Union (AOU). 1957. Check- 
list of North American birds. 5th Ed. American Or- 
nithologists’ Union, Washington, DC U.S.A. 
Anderson, D.R., K.P. Burnham, A.B. Franklin, RJ. Gu- 
tierrez, E.D. Forsman, R.G. Anthony, G.G. White, 
AND T.M. Shenk. 1999. A protocol for conflict reso- 
lution in analyzing empirical data related to natural 
resource controversies. Wildl. Soc, Bull, 27:1050-1058. 
Austin, K.K. 1993. Habitat use and home range size of 
breeding Northern Goshawks in the southern Cas- 
cades. M.S. thesis, Oregon State Univ., Corvallis, OR 
U.S.A. 

Baird, S.F., T.M. Brewer, and R. Ridgway. 1875. A history 
of North American birds. Vol. 3. Land birds. Little, 
Brown, Boston, MA U.S.A. 

Bednarz, J.C., D. Klem, Jr., LJ. Goodrich, and S.E. Sen- 
NER. 1990. Migration counts of raptors at Hawk Moun- 
tain, Pennsylvania, as indicators of population trends, 
1934-86. Auk 107:96-109. 

Beier, P. 1997. Winter foraging habitat of Northern Gos- 
hawks in northern Arizona. Final report. Arizona 
Game and Fish Department Heritage Grant 195032, 
Pheonix, AZ U.S.A. 

and J.E. Drennan. 1997. Forest structure and 

prey abundance in foraging areas of Northern Gos- 
hawks. Ecol. Appl. 7:564-571. 

Bildstein, K.L. 1998. Long-term counts of migrating rap- 
tors: a role for volunteers in wildlife research./. Wildl. 
Manag. 62:435-445. 

Boal, C.W. and R.W. Mannan. 1994. Northern Goshawk 
diets in ponderosa pine forests on the Kaibab Plateau. 
Stud. Avian Biol. 16:97-102. 

Bosakowski, T. 1999. The Northern Goshawk: ecology, 
behavior, and management in North America. Han- 
cock House, Blaine, WA U.S.A. 

Braun, C.E., J.H. Enderson, M.R. Fuller, Y.B. Linhart, 
AND C.D. Marti. 1996. Northern Goshawk and forest 
management in the southwestern United States. The 
Wildlife Society, Bethesda, MD U.S.A. 

Bright-Smith, D.J. and R.W. Mannan. 1994. Habitat use 
by breeding male Northern Goshawks in northern Ar- 
izona. Stud. Avian Biol. 16:58—65. 

Crocker-Bedford, D.C. 1990. Goshawk reproduction 
and forest management. Wildl. Soc. Bull. 18:262-269. 

. 1998. The value of demographic and habitat 

studies in determining the status of Northern Gos- 


hawks {Accipiter gentilis atricapillus) with special refer- 
ence to Crocker-Bedford (1990) and Kennedy (1997). 
J. Raptor Res. 32:329-336. 

AND B. Chaney. 1988. Characteristics of goshawk 

nesting stands. Pages 210-217 in R. Glinski, B.G. Pen- 
dleton, M.B. Moss, B.A. Millsap, and S.W. Hoffman 
[Eds.], Southwest raptor management symposium 
and workshop. Natl. Wildl. Fed. Sci. Tech. Ser. 11. 

Daw, S.K. 1997. Northern Goshawk nest site selection 
and habitat associations at the post-fledging family 
area scale in Oregon. M.S. thesis, Oregon State Univ., 
Corvallis, OR U.S.A. 

and S. DeStefano. 2001. Forest characteristics of 

Northern Goshawk nest stands and post-fledging ar- 
eas in Oregon. J. Wildl. Manag. 65:59-65. 

, S. DeStefano, and RJ. Steidl. 1998. Does survey 

method bias the description of Northern Goshawk 
nest-site structure? J. Wildl. Manag. 62:1379-1384. 

Desimone, S.M. 1997. Occupancy rates and habitat rela- 
tionships of Northern Goshawks in historic nesting 
areas in Oregon. M.S. thesis, Oregon State Univ., Cor- 
vallis, OR U.S.A. 

DeStefano, S. 1998. Determining the status of Northern 
Goshawks in the west: is our conceptual model cor- 
rect? /. Raptor Res. 32:342-348. 

, S.K. Daw, S.M. Desimone, and E.C. Meslow 

1994a. Density and productivity of Northern Gos- 
hawks: implications for monitoring and management. 
Stud. Avian Biol. 16:88-91. 

, B. Woodbridge, and P. Detrich. 1994b. Survival 

of Northern Goshawks in the southern Cascades of 
California. Stud. Avian Biol. 16:133-136. 

Doyle, F.I. and J.M.N. Smith. 1994. Population responses 
of Northern Goshawks to the 10-year cycle in num- 
bers of snowshoe hares. Stud. Avian Biol. 16:122-129. 

and . 2001. Raptors and scavengers. Pages 

377-404 in C.J. Krebs, S. Boutin, and R. Boonstra 
[Eds.], Ecosystem dynamics of the boreal forest: the 
Kluane Project. Oxford Univ. Press, New York, NY 
U.S.A. 

Drennan, J.E, and P. Beier. 2003. Forest structure and 
prey abundance in winter habitat of Northern Gos- 
hawks./. Wildl. Manag. 67:177-185. 

Finn, S.P. 2000. Multi-scale habitat influences on North- 
ern Goshawk occupancy and reproduction on the 
Olympic Peninsula, Washington. M.S. thesis, Boise 
State Univ., Boise, ID U.S.A. 

, J.M. Marzluff, and D.E. Varland. 2002. Effects 

of landscape and local habitat attributes on Northern 
Goshawk site occupancy in western Washington. For 
Sci. 48:427-436. 

Flatten, C., K. Titus, and R. Lowell. 2001. Northern 
Goshawk monitoring, population ecology and diet on 
the Tongass National Forest: annual monitoring re- 
port. Alaska Department of Fish and Game, Juneau, 
AK U.S.A. 

Franklin, A.B. 1992. Population regulation in northern 


206 


Andersen et al. 


VoL. 39, No. 3 


Spotted Owls: theoretical implications for manage- 
ment. Pages 815-827 in D.R. McCullough and R.H. 
Barrett [Eds.], Wildlife 2001: populations. Elsevier 
Applied Science, London, U.K. 

Gavin, T.A. and B. May. 1996. Genetic variation and tax- 
onomic status of Northern Goshawks in Arizona: im- 
plications for management. Unpublished report sub- 
mitted to Arizona Game and Fish Department. 
Cornell Univ., Ithaca, NY U.S.A. 

Graham, R.T., R.T. Reynolds, M.H. Reiser, R.L. Bassett, 
AND D.A. Boyce. 1994. Sustaining forest habitat for 
the Northern Goshawk; a question of scale. Stud. Avi- 
an Biol. 16:12-17. 

Grinnell, J. AND A.H. Miller. 1944. The distribution of 
the birds of California. Pacific Coast Avifauna 27. 

Hall, L.S., P.R. Krausman, and M.L. Morrison. 1997. 
The habitat concept and a plea for standard termi- 
nology. Wildl. Soc. Bull. 25:173-182. 

Hargis, C.D., C. McCarthy, and R.D. Perloff. 1994. 
Home ranges and habitats of Northern Goshawk in 
eastern California. Stud. Avian Biol. 16:66-74. 

Hayward, G.D. and R.E. Escano. 1989. Goshawk nest site 
characteristics in western Montana and northern Ida- 
ho. ConcJor 91:476-479. 

Hunt, W.G. 1998. Raptor floaters at Moffat’s equilibrium. 
Oikos 82:191-197. 

Ingraldi, M.F. 1998. Population biology of Northern 
Goshawks in east-central Arizona. Nongame and En- 
dangered Wildlife Program Technical Report 133. Ar- 
izona Game and Fish Department, Phoenix, AZ 
U.S.A. 

Iverson, G.C., G.D. Hayward, K. Titus, E. DeGavner, 
R.E. Lowell, D.C. Crocker-Bedford, P.F. Schempf, 
andJ. Lindell. 1996. Conservation assessment for the 
Northern Goshawk in southeast Alaska. General Tech- 
nical Report PNW-GTR-387. U.S. Department of Ag- 
riculture, Forest Service, Pacific Northwest Research 
Station, Portland, OR U.S.A. 

Johansson, C., P.J. Hardin, and C.M. White. 1994. 
Large-area goshawk habitat modeling in Dixie Nation- 
al Forest using vegetation and elevation data. Stud. 
Avian Biol. 16:50-57. 

Joy, S.M. 2002. Northern Goshawk habitat on the Kaibab 
National Forest in Arizona: factors affecting nest lo- 
cations and territory quality. Ph.D. dissertation, Col- 
orado State University, Fort Collins, CO U.S.A. 

, R.T. Reynolds, and D.G. Leslie. 1994. Northern 

Goshawk broadcast surveys: hawk response variables 
and survey cost. Stud. Avian Biol. 16:24-31. 

Keane, J.J. 1999. Ecology of the Northern Goshawk in the 
Sierra Nevada, California. Ph.D. dissertation, Univ. 
California, Davis, CA U.S.A. 

Kennedy, P.L. 1991. Reproductive strategies of Northern 
Goshawks and Cooper’s Hawks in north-central New 
Mexico. Ph.D. dissertation, Utah State Univ., Logan, 
UT U.S.A. 

. 1997. The Northern Goshawk {Accipiter gentilis 


atricapillus) : is there evidence of a population decline? 
J. Raptor Res. 31:95-106. 

. 1998. Evaluating Northern Goshawk {Accipiter 

gentilis atricapillus) population trends: a reply to Small- 
wood and Crocker-Bedford. J. Raptor Res. 32:336-342. 

AND D.W. Stahlecker. 1993. Responsiveness of 

nesting Northern Goshawks to taped broadcasts of 
three conspecific calls./. Wildl. Manag. 57:249-257. 

Kostrzewa, a. and R. Kostrzewa. 1990. The relationship 
of spring and summer weather with density and 
breeding performance of the buzzard Buteo buteo, gos- 
hawk Accipiter gentilis and kestrel Falco tinnunculus. Ibis 
132:550-559. 

Lewis, S.B. 2001. Breeding season diet of Northern Gos- 
hawks in southeast Alaska with a comparison of tech- 
niques used to examine raptor diet. M.S. thesis, Boise 
State Univ., ID U.S.A. 

Lilieholm, RJ. J .N. Long, and S. Patia. 1994. Assess- 
ment of goshawk nest area habitat using stand density 
index. Stud. Avian Biol. 16:18-23. 

Long, J.N. and F.W. Smith. 2000. Restructuring the for- 
est: goshawks and the restoration of southwestern 
ponderosa pine. J. For. 98:25-30. 

Maguire, L. and D. Call. 1993. Population viability anal- 
ysis of Northern Goshawks {Accipiter gentilis) on the 
North Kaibab Ranger District, Arizona. Contract No. 
G10076-A. Arizona Game and Fish Department, Phoe- 
nix, AZ U.S.A. 

Maurer, J.R. 2000. Nesting habitat and prey relations of 
the Northern Goshawk in Yosemite National Park, 
California. M.S. thesis, Univ. California, Davis, CA 
U.S.A. 

McClaren, E.L., P.L. Kennedy, and S.R. Dewey. 2002. Do 
some Northern Goshawk nest areas consistently 
fledge more young than others? Condor 104:343-352. 

McGrath, M.T. 1997. Northern Goshawk habitat analysis 
in managed forest landscapes. M.S. thesis, Oregon 
State Univ., Corvallis, OR U.S.A. 

, S. DeStefano, R.A. Riggs, L.L. Irwin, and G.J. 

Roloff. 2003. Spatially explicit influences on North- 
ern Goshawk nesting habitat in the interior Pacific 
Northwest. Wildl. Monogr. 154:1-63. 

Meehan, T.C., C.A. Lott, Z.D. Sharp, R.B. Smith, R.N. 
Rosenfield, A.C. Stewart, and R.K. Murphy. 2001. 
Using hydrogen isotope geochemistry to estimate the 
natal latitudes of immature Cooper’s Hawks migrating 
through the Florida Keys. Condor 103:11—20. 

Moore, K.R. and C.J. Henny. 1983. Nest site character- 
istics of three coexisting accipiter hawks in northeast- 
ern Oregon. Raptor Res. 17:65-76. 

Mueller, H.C., D.D. Berger, and G. Allez. 1976. Age 
and sex variation in the size of goshawks. Bird-Banding 
47:310-318. 

, , and . 1977. The periodic invasions 

of goshawks. Auk 94:652-663. 

Newton, I. 1989. Sparrowhawk. Pages 279-296 ini. New- 


September 2005 


Status 


207 


ton [Ed.], Lifetime reproduction in birds. Academic 
Press, London, U.K. 

. 1991. Habitat variation and population regula- 
tion in sparrowhawks. Ibis 133 (Supplement I):76-88. 

Patla, S.M. 1997. Nesting ecology and habitat of the 
Northern Goshawk in undisturbed and timber harvest 
areas on the Targhee National Forest, Greater Yellow- 
stone Ecosystem. M.S. thesis, Idaho State Univ., Po- 
catello, ID U.S.A. 

Pendleton, G.W., K. Titus, E. DeGavner, C.J. Flatten, 
AND R.E. Lowell. 1998. Compositional analysis and 
GIS for study of habitat selection by goshawks in 
Southeast Alaska. J. Agric. Biol. Environ. Stat. 3:280— 
295. 

POSTUPAI.SKY, S. 1974. Raptor reproductive success: some 
problems with methods, criteria, and terminology. 
Pages 21—31 mF.N. Hamerstrom Jr., B.E. Harrell, and 
R.R. Olendorff [Eds]. Raptor Research Report No. 2. 
Management of raptors. Raptor Research Founda- 
tion, Inc., Vermillion, SD U.S.A. 

Pradel, R. 1996. Utilization of capture-mark-recapture 
for the study of recruitment and population growth 
rate. Biometrics 52:703-709. 

Reich, R.M., S.M. Joy, and R.T. Reynolds. 2004. Predict- 
ing the location of Northern Goshawk nests: model- 
ing the spatial dependency between nest locations 
and forest structure. Ecol. Model. 176:109-133. 

Reynolds, R.T. 1989. Accipiters. Pages 92-101 in B.A. 
Pendleton, C.E. Ruibal, D.L. Krahe, K. Steenhof, M.N. 
Kockert, and M.N. LeFranc, Jr. [Eds.], Proceedings of 
the western raptor management symposium and 
workshop. National Wildlife Federation Scientific and 
Technical Series No. 12. 

, W.M. Block, and D.A. Boyce, Jr. 1996. Using 

ecological relationships of wildlife as templates for re- 
storing southwestern forests. Pages 35-43 in W. Cov- 
ington and P.K. Wagner, technical coordinators. Con- 
ference on adaptive ecosystem restoration and 
management: restoration of Cordilleran conifer land- 
scapes in North America. General Technical Report 
RM-GTR-278. U.S. Department of Agriculture, Forest 
Service, Rocky Mountain Research Station, Fort Col- 
lins, CO U.S.A. 

, R.T. Graham, M.H. Reiser, R.L. Bassett, P.L. 

Kennedy, D.A. Boyce, Jr., G. Goodwin, R. Smith, and 
E.L. Fisher. 1992. Management recommendations for 
the Northern Goshawk in the southwestern United 
States. General Technical Report RM-217. U.S. De- 
partment of Agriculture, Forest Service, Rocky Moun- 
tain Forest and Range Experiment Station, Fort Col- 
lins, CO U.S.A. 

and S.M. Joy. 1998. Distribution, territory occu- 
pancy, dispersal, and demography of Northern Gos- 
hawks on the Kaibab Plateau, Arizona. Final report 
for Heritage Project No. 194045. Arizona Game and 
Fish Department, Phoenix, AZ U.S. A. 

and E.C. Meslow. 1984. Partitioning of food and 


niche characteristics of coexisting Accipiter during 
breeding. Auk 101:761-779. 

, E.C. Meslow, and H.M. Wight. 1982. Nesting 

habitat of coexisting Accipiter in Oregon. /. Wildl, Man- 
ag. 46:124-138. 

Shaffer, H.B., R.N. Fisher, and C. Davidson. 1998. The 
role of natural history collections in documenting 
species declines. Trends Ecol. Evol. 13:27—30. 

Smallwood, K.S. 1998. On the evidence needed for list- 
ing Northern Goshawks {Accipiter gentilis) under the 
Endangered Species Act: a reply to Kennedy. J. Raptor 
Res. 32:323-329. 

SONSTHAGEN, S.A., S.L. TALBOT, AND C.M. WHITE. 2004. 
Gene flow and genetic characterization of Northern 
Goshawks breeding in Utah. Condor 106:826— 836. 

Squires, J.R. and L.F. Ruggiero. 1995. Winter move- 
ments of adult Northern Goshawks that nested in 
southcentral Wyoming. J. Raptor Res. 29:5-8. 

AND R.T. Reynolds. 1997. Northern Goshawk {Ac- 
cipiter gentilis) . In A. Poole and F. Gill [Eds.], The birds 
of North America, No. 298. The Birds of North Amer- 
ica, Inc., Philadelphia, PA U.S. A. 

Stresemann, E. and D. Amadon. 1979. Falconiformes 
Pages 270-425 in E. Mayr and G.W. Cottrell [Eds.], 
Check-list of birds of the world: revision of the work 
of James L. Peters. Museum of Comparative Zoology, 
Cambridge, MA U.S.A. 

Taverner, P.A, 1940. Variation in the American goshawk 
Condor 42:157-160. 

Trexel, D.R., R.N. Rosenfield, J. Bielefeldt, and E.A. 
Jacobs. 1999. Comparative nest site habitats in Sharp- 
shinned and Cooper’s hawks in Wisconsin. Wilson 
Bull. 111:7-14. 

United States Department of Interior (USDI). 1997. 
Endangered and threatened wildlife and plants: 90- 
day finding for a petition to list the Northern Gos- 
hawk in the contiguous United States west of the 100* 
meridian. Federal Register 62:50892—50896. 

. 1998. Endangered and threatened wildlife and 

plants: notice of 12-month finding on a petition to list 
the Northern Goshawk in the contiguous United 
States west of the 100* meridian. Federal Register 63. 
35183-35184. 

and United States Department of Commerce. 

1996. Policy regarding the recognition of distinct ver- 
tebrate population segments under the Endangered 
Species Act. Federal Register 61:4721-4725. 

United States Fish and Wildlife Service (FWS). 1998. 
Northern Goshawk status review. U.S. Fish and Wild- 
life Service, Office of Technical Support, Portland, 
OR, U.S.A. 

Ward, L.Z., D.K. Ward, and TJ. Tibbits. 1992. Canopy 
density analysis at goshawk nesting territories on the 
North Kaibab Ranger District, Kaibab National Forest: 
final report. Arizona Game and Fish Department, 
Phoenix, AZ U.S.A. 

Warren, N.M., G.D. Hayward, R. Holland, R. Escano, 


208 


Andersen et al. 


VoL. 39, No. 3 


D.C. Crocker-Bedford, T. Komberec, D. Sasse, L. 
Saunders-Ogg, and W.C. Shuster. 1990. Goshawk 
habitat relationships. Pages 12—27 in N.M. Warren 
[Ed.], Old-growth habitats and associated wildlife spe- 
cies in the northern Rocky Mountains. Publication 
Rl-90-42. U.S. Department of Agriculture, Forest Ser- 
vice, Northern Region, Missoula, MT U.S.A. 

Watson, J.W., D.W. Havs, S.P. Finn, and P. Meehan-Mar- 
TIN. 1998. Prey of breeding Northern Goshawks in 
Washington./. Raptor Res. 32:297-305. 

Whaley, W.H. and C.M. White. 1994. Trends in geo- 
graphic variation of Cooper’s Hawk and Northern 
Goshawk in North America: a multivariate analysis. 
Proc. West. Found. Vertebr. Zool. 5:161-209. 

Widen, P. 1989. The hunting habitats of goshawks Acdp- 
iter gentilis in boreal forests of central Sweden. 131: 
205-213. 

. 1997. How, and why, is the goshawk {Accipiter gen- 
tilis) affected by modern forest management in Fen- 
noscandia? J. Raptor Res. 31:107-113. 

WOODBRIDGE, B. AND P.J. Detrich. 1994. Territory occu- 
pancy and habitat patch size of Northern Goshawks 
in the southern Cascades of California. Stud. Avian 
Biol. 16:83-87. 

Younk, J.V. AND M.J. Bechard. 1994. Breeding ecology of 
the Northern Goshawk in high-elevation aspen forests 
of northern Nevada. Stud. Avian Biol. 16:119-121. 

Received 30 January 2004; accepted 28 September 2004 

Guest Editor: Patricia L. Kennedy 

Appendix. Definitions of ecological terms as they apply 

to Northern Goshawks. 

“Active” nest: The term “active” as applied to raptor nests 
was hrst dehned by Postupalsky (1974) to describe a 
nest where si egg was laid. However, the term has 
been used in different ways since then, and is probably 
best avoided. 

Breeding area: a nesting area used by goshawks in the pres- 
ent, past, or both. 

Breeding area occupancy: goshawks are thought to defend 
use areas from conspecifics (territories) during the 
breeding season, and these territories are often used 
m subsequent years. However, because it is generally 
impractical to assess territory occupancy, occupancy of 
breeding areas has been assessed in field studies of gos- 
hawks. Breeding areas are occupied when goshawks are 
present, and what constitutes presence has been vari- 
able across studies, or is undefined. We suggest that 
breeding areas are occupied when any of the following 
occur: (1) nesting, (2) one or more goshawks are ob- 
served in association with a nest with evidence of 
recent use (e.g., fresh greenery or other evidence of 
recent nest construction), (3) goshawks respond ag- 
gressively to humans or respond to conspecific call 
broadcasts during the breeding season, or (4) pre-dis- 
persal fledglings are located in the vicinity of a nest 
that has evidence of recently being used (e.g., fresh 


whitewash, goshawk feathers, prey remains, or pellets) . 
If none of these conditions exist, a breeding area can- 
not be assumed to be unoccupied without meeting 
additional criteria (e.g., no goshawk detection during 
systematic searching for nests or in response to con- 
specific call broadcasts) . Consistent, specific criteria for 
categorizing a breeding area as unoccupied need to be 
developed. 

Breeding density: the number of nests used by breeding 
goshawks per unit area; alternatively, the number of 
goshawk breeding areas through a specified time pe- 
riod per unit area. 

Breeding population: a group of goshawks that interact in 
space and time and that breed or potentially breed and 
for which it is reasonable to discuss emergent popula- 
tion properties, such as rate of growth, productivity, 
etc. Goshawk populations are delimited by spatial 
boundaries based on where they breed, but these 
boundaries may not be relevant throughout an annual 
period (e.g., goshawks that annually migrate from 
breeding areas) or from one year to the next (e.g., 
goshawks that migrate from breeding areas in only 
some years). 

Habitat: the collection of biotic and abiotic factors that 
produce occupancy by goshawks {sensu Hall et al. 
1997). 

Nest(ing) area: the immediate area surrounding (a) 
nest(s) used by breeding goshawks. 

Nest(ing) attempt: a nest that has been used in any manner 
by goshawks during the breeding season. Goshawks 
can be observed at a nest, or there may be evidence of 
egg laying (e.g., eggs or egg fragments), nestlings, or 
fledglings. Other evidence is often used to iiifer that 
an egg has been laid or that a pair of goshawks is pre- 
paring to lay eggs, including observation of goshawks 
reconstructing an existing nest or building a new nest, 
observation of greenery added to existing nests, pres- 
ence of recently molted goshawk feathers in or be- 
neath a nest, etc. A nest attempt does not necessarily 
result in egg laying (i.e., nest failure can occur prior 
to egg laying). 

Nest stand: the area covered by a forested patch consisting 
of trees that are often characterized by having a similar 
size, species, and spacing and in which a goshawk nest 
occurs. 

Nest(ing) success: the proportion of nests in which eggs are 
laid that produce at least 1 fledgling. 

Nest tree: the tree in which a goshawk nest is placed. 

Occupied nest area: an area on which a pair of goshawks 
have established residency during the nesting season 
and includes ^1 nest. 

Post-fledging area: the area that is used by recently fledged 
goshawks before they become independent of adults 
(sensu Reynolds et al. 1992). 

Successful nest: a nesting attempt that results in S;1 young 
fledged. 

Territory: an area defended by goshawks from conspecifics 
during the breeding season that generally contains the 


September 2005 


Status 


209 


nest, alternative nest(s), if any, nest stand(s), nesting 
area, post-fledging area, and at least some of the area 
used by adults for foraging. 

Use area-home range: area traversed by a goshawk or pair 
of goshawks during the course of normal, daily activi- 


ties. It is generally necessary to define specific time 
periods over which use areas or home ranges apply, as 
they can change in size and other attributes through 
time. 



Buteq 

Books 


Toll Free: 800-722-2460 
phone: 434-263-8671 
fax: 434-263-4842 


Specializing in Ornithology 


Buteo Books is the largest retailer of Ornithology books in North America, 
with over 2,000 in-print titles, arid hundreds of out-of-print titles available. 




WAfCM 


UNPtfiMrtNOlKf 

OWLS 



NEW, IN-PRINT ORNITHOLOGY 

The most popular authors, cutting edge studies, and unusual subjects. We stock scientific 
texts, hard-to-find foreign books, and entertaining reading; plus audio, video, and software. 

RARE AND OUT-OF-PRINT ORNITHOLOGY 

Buteo Books has a wide selection of used, out-of-print, and scarce titles available, including 
classics on birds of prey and falconry. Our stock changes weekly, so call to check availability. 


BIRDS OF NORTH AMERICA SERIES 



This series consists of individual species accounts for each of the 
species which breed in the United States and Canada, including 
diurnal and nocturnal raptors. These illustrated reviews provide 
comprehensive summaries of the current knowledge of each species, 
with range maps and extensive lists of references. 

As the exclusive distributor of print copies in this series, Buteo Books 
is pleased to offer these accounts for $7.50 each. All 716 profiles are 
listed in taxonomic order on our website. 


Buteo Books; 3130 Laurel Road; Shipman, VA 22971; USA 


Visit our website for more information; 

www.buteobooks.com 






J Raptor Res. 39(3);210-221 
© 2005 The Raptor Research Foundation, Inc. 


IS FLEDGING SUCCESS A RELIABLE INDEX OE EITNESS 

IN NORTHERN GOSHAWKS? 


J. David Wiens^ and Richard T. Reynolds 

USD A Forest Service, Rocky Mountain Research Station, 2150 Centre Avenue, Building A, Suite 350, 

Fort Collins, CO 80526 US. A. 


Abstract. — Fledging success is often assumed to be a reliable index of reproductive success (i.e., fitness) 
in the ornithological literature. We examined the validity of this assumption in a large population of 
Northern Goshawks {Accipiter gentilis) on the Kaibab Plateau, Arizona, at both the population and in- 
dividual levels. We used mark-recapture data from 558 fledglings produced at 494 nests over a 10-yr 
period to assess the hypothesis that the number of fledglings returning to breed from an annual fledg- 
ling cohort is positively correlated with the size of the cohort. Natal philopatry was low and recruitment 
was gradual: only 48 fledglings (8.6%) returned to breed between 2-8 yr of age (x = 3.5 yr). We found 
no evidence that the breeding population produced more local recruits in years of high fledgling 
production than in years of low fledgling production. At the individual level, however, fledgling pro- 
duction for 290 breeding adults was related to their contributions to the future breeding population. 
Variation in fitness potential among territorial adults was high, as only 20% of the breeding population 
produced nearly 50% of the fledglings and 84% of the local recruits during the study. Our results 
indicate that measures of annual productivity for a large breeding population were not reflective of 
reproductive success, whereas measures of individual productivity were. We conclude that fledging suc- 
cess of individual goshawks is a reliable index of fitness, but that population productivity is a poor 
predictor of local recruitment. 

Keywords: Northern Goshawk, Accipiter gentilis; individual heterogeneity, Kaibab Plateau', recruitment; 

reproductive success. 


iCONSTITUYE EL EXITO DE EMPLUMAMIENTO UN INDICE CONFIABLE DE ADECUACION 
BIOLOGICA EN ACCIPITER GENTILIS? 

Resumen. — En la literatura ornitologica, a menudo se asume que el exito de emplumamiento representa 
un indice confiable del exito reproductivo (i.e., adecuacion biologica) . Examinamos la validez de este 
supuesto en una poblacion de gran tamaho de Accipiter gentilis en Kaibab Plateau, Arizona, tan to a nivel 
de poblacion como de individuo. Usamos datos de marcado y recaptura de 558 volantones provenientes 
de 494 nidos a lo largo de un periodo de 10 ahos para evaluar la hipotesis de que el numero de 
volantones que regresan a reproducirse de una cohorte anual de volantones esta positivamente corre- 
lacionado con el tamaho de la cohorte. La filopatria natal fue baja y el reclutamiento fue gradual: solo 
48 volantones (8.6%) regresaron a reproducirse con entre dos y ocho ahos de edad (x = 3.5 ahos). No 
encontramos evidencia de que la poblacion reproductiva produjo mas reclutamientos locales en ahos 
de alta produccion de volantones que en ahos de baja produccion. A nivel individual, sin embargo, la 
produccion de volantones correspondiente a 290 adultos reproductivos se relaciono con sus contribu- 
ciones a la futura poblacion reproductiva. La variacion en la adecuacion biologica entre los territorios 
de los adultos fue alta, ya que solo el 20% de la poblacion reproductiva produjo cerca del 50% de los 
volantones y el 84% de los reclutamientos locales durante el estudio. Nuestros resultados indican que 
las medidas de productividad anual de una poblacion reproductiva de gran tamaho no reflejaron el 
exito reproductivo, mientras que las medidas de productividad individual si lo hacen. Concluimos que 
el exito de emplumamiento individual de A. gentilis representa un indice confiable de adecuacion biol- 
ogica, pero que la productividad a nivel poblacional predice de modo inadecuado el reclutamiento 
local. 

[Traduccion del equipo editorial] 


^ Gorresponding author’s email address: jdwiens@comcast.net 


210 


September 2005 


Biology 


211 


In birds, a commonly measured reproductive 
variable used to assess population performance 
over time is fledging success (the number of young 
that fledge per nest) . A widely held assumption in 
avian studies is that fledging success is a reliable 
index of reproductive success (the number of off- 
spring that survive to become breeding adults) and 
thus, fitness (Endler 1986, Weatherhead and Du- 
four 2000, Keedwell 2003). However, the spatial 
and temporal scales over which many populations 
are studied may not correspond well with the spa- 
tial extent of natal dispersal or the time periods 
over which recruitment occurs. These limitations, 
superimposed on a variety of stochastic factors, 
could easily disrupt the relationship between fledg- 
ing success and fitness. Nevertheless, researchers 
often assume this relationship is positive, in part, 
due to the difficulties associated with estimating 
pre-breeding survival and emigration rates, even in 
long-term banding studies. Difficulties in attaining 
direct measures of reproductive success are there- 
fore exemplified in long-lived, wide-ranging spe- 
cies that occur at low densities, have low natal phil- 
opatry, elude detection when not breeding, and 
initiate breeding at a delayed age (Weatherhead 
and Dufour 2000). 

The Northern Goshawk (Accipiter gentilis) is a 
long-lived raptor that occupies mature temperate 
and boreal forests throughout the Holarctic 
(Squires and Reynolds 1997). The goshawk is a so- 
cially monogamous, territorial species that lays one 
clutch per year (Reynolds et al. 1994, Kennedy and 
Ward 2003). Although several studies have docu- 
mented extensive temporal variation in fledging 
success in goshawk populations (e.g., McClaren et 
al. 2002, Reynolds and Joy in press), none have 
addressed the relationship between fledging suc- 
cess and local recruitment. The strength of this re- 
lationship is particularly relevant when the dynam- 
ics of a local population are more heavily reliant 
on external recruitment (i.e., immigration) than 
internal productivity, which may occur when natal 
philopatry is low and adult site fidelity is high (Sta- 
cey and Taper 1992, Martin et al. 2000). While 
mate and site fidelity in adult goshawks is high 
(75-95%; Detrich and Woodbridge 1994, Reynolds 
and Joy in press) , the degree of natal philopatry is 
largely unknown due to low juvenile recapture 
probabilities, few recoveries of banded nestlings, 
and a general lack of information on the extent of 
natal dispersal (Kennedy and Ward 2003, but see 
Wiens 2004). However, molecular evidence has 


shown that gene flow among subpopulations of 
goshawks over large geographic areas is high (Son- 
sthagen 2004, Bayard de Volo et al. 2005), indicat- 
ing that juveniles may disperse over long distances 
because adults rarely disperse once they have set- 
tled on a breeding territory. Juvenile survival is one 
of the most difficult demographic parameters to 
estimate precisely in goshawks (Kennedy 1997, 
Wiens 2004), further emphasizing the need to as- 
sess the assumed relationship between fledging 
production and reproductive success. 

In this paper we evaluate whether fledgling suc- 
cess is a reliable predictor of reproductive success 
in goshawks. We examined this relationship at both 
the population and individual levels using a 13-yr 
mark-recapture (resight) data set obtained from a 
breeding population of goshawks exceeding 120 
pairs on the Kaibab Plateau, Arizona. Our investi- 
gation was inspired by studies showing a positive 
relationship between fledging success and recruit- 
ment in bird species such as Red-winged Blackbirds 
(Agelaius phoeniceus) , Eurasian Sparrowhawks {Ac- 
cipiter nisus) , Ural Owls {Strix uralensis) , and Osprey 
{Pandion haliaetus; Weatherhead and Dufour 2000, 
Newton 1989a, Saurola 1989, Postupalsky 1989, re- 
spectively) . At the population level, we assessed the 
hypothesis that the number of fledglings produced 
annually is positively correlated with the number 
of individuals from annual cohorts that were even- 
tually recruited into the local breeding population. 
At the individual level, we anticipated that total 
fledgling production of color-marked male and fe- 
male adult (>2 yr old) goshawks would be posi- 
tively related to the number of their descendants 
that were recruited into the local breeding popu- 
lation. In examining our hypotheses, we also re- 
port on local recruitment and ages at first breeding 
for goshawks on the Kaibab Plateau. 

Methods 

Study Area and Field Procedures. The study area in- 
cluded all the coniferous forest above 2182 m elevation 
on the Kaibab Plateau of northern Arizona. This 1732 
km^ area included the northern portion of the Kaibab 
National Forest and the Grand Canyon National Park 
(North Rim). The Kaibab Plateau is a large (95 X 55 
km), oval-shaped plateau that rises from a shrub-steppe 
plain at 1750 m elevation to the highest point at 2800 m 
and is dissected by moderately sloping valleys (Rasmussen 
1941). Forests of the Kaibab Plateau are isolated from 
similar forests by variable distances (35-250 km) of pin- 
yon-juniper {Pinus edulis-Juniperus spp.) woodland, and 
sagebrush {Artemisia spp.) plains. See Reynolds et al. 
(1994) and Reynolds et al. (2005) for further detail on 
the study area, its management history, and protocols 


212 


Wiens and Reynolds 


VoL. 39, No. 3 


used to locate, survey, and monitor goshawk breeding 
areas. 

We defined a territory as a breeding area used, but not 
necessarily defended against conspecifics, by a single pair 
of goshawks during a breeding season (Reynolds et al. 
2005). During 1991-2003, a high density of regularly- 
spaced goshawk breeding territories were identified on 
the Kaibab Plateau (Reich et al. 2004, Reynolds et al. 
2005) . Territories contained multiple alternate nests that 
were used one or more times over the years by goshawks. 
We visited all nests in early spring of each year to estimate 
occupancy and reproductive status of territorial pairs. If 
goshawks were not using a known nest, a three-stage ter- 
ritory survey protocol (Reynolds et al. 2005) was used to 
determine territory status. We classified a territory as "ac- 
tive' when eggs were laid, “occupied' when adult gos- 
hawks were observed on two or more occasions in the 
vicinity of a nest (or a single observation of an adult in 
combination with observations of molted feathers, feces, 
and fresh nest construction), and “unknown" if eggs were 
not laid and no evidence of goshawk occupancy was 
found (Reynolds et al. 2005). We determined nest fate 
(successful = fledged ^1 young; failed = eggs laid but 
no fledglings produced), nest productivity (number of 
fledglings) , and identity of adults during weekly visits to 
active territories. For our purposes here, we defined the 
number of young fledged as the number of nestlings 
present at the time of banding (ca. 1 wk prior to fledg- 
ing) . Nesting adults were captured, measured, sexed, and 
aged during the mid-late nestling period following meth- 
ods described in Reynolds et al. (1994); nestlings were 
captured by climbing nest trees during the last week of 
the nestling period (mid-late June). Sex of nestlings was 
assigned on the basis of body mass, tarsometatarsal 
length, and toepad-length measurements (Wiens 2004). 
All captured hawks received a U.S. Geological Survey alu- 
minum leg band and an anodized colored leg band with 
a unique alpha-numeric code (Arcraft Sign and Plate Co., 
Edmonton, Canada) readable to 80 m with 40-60 X spot- 
ting scopes. 

Age at First Breeding and Local Recruitment, We con- 
sidered a goshawk to have been recruited locally if it was 
banded as a nestling on the study area and later observed 
breeding in the study population. Thus, hawks classified 
as “local recruits” had to have been recaptured or re- 
sighted at an active territory on the Kaibab Plateau. To 
attain an unbiased estimate of the age at first breeding, 
we included only banded (known-aged) hawks that we 
were confident had been detected on their first breeding 
attempt. This meant that banded recruits had to have 
been resighted on territories where the same-sex occu- 
pant during the prior year was known. Estimated ages at 
first breeding could have been biased high by breeding 
dispersal (i.e., undetected movement among territories 
between successive breeding attempts). However, this 
bias was likely to be small because <6% of adult goshawks 
moved to a different territory between successive breed- 
ing attempts (R. Reynolds unpubl. data). To attain a 
mean estimate of local recruitment, we subtracted the 
number of nestlings that were too young (based on the 
median age at first breeding) to have attained breeding 
territories by spring 2003 from the total number of nest- 
lings banded during 1991-2003. Hence, local recruit- 


ment was calculated as the total number of banded re- 
cruits detected during 1991-2003 divided by the total 
number of nestlings banded during 1991-2000. 

Population Productivity. We were unable to capture 
and mark all nestlings produced in some years because 
several nests were not located until after fledging or, rare- 
ly, a nest tree was unsafe to climb. Thus, the number of 
young banded represented a portion (69% during 1991- 
2003) of the known number of young that actually 
fledged. For this reason, we defined “productivity” as the 
number of young banded. We assumed that banded 
young comprised a representative sample in terms of 
population productivity and local recruitment. Using the 
number of young banded annually, we assessed whether 
the size of each annual fledgling cohort was correlated 
with the number of local recruits originating from each 
cohort. We conducted analyses with sexes combined and 
then separately for males and females. Data were exam- 
ined by sex for two reasons. First, patterns of post-fledg- 
ing mortality have been found to differ for male and fe- 
male goshawks on the Kaibab Plateau (Wiens 2004). 
Second, natal dispersal distances may differ between sex- 
es (Greenwood and Harvey 1982, Byholm et al. 2003). 
One or both of these factors could result in sex-depen- 
dent local recruitment. For sex-specific analyses, the 
number of individuals of each sex that successfully re- 
cruited from a fledgling cohort was assessed relative to 
the total number of individuals of each sex that were 
banded as nestlings. 

Individual Productivity. To examine the relationship 
between total fledgling production of individual adult 
goshawks and recruitment success of their offspring, we 
calculated the number of banded fledglings produced by 
each color-marked adult that bred during 1991-2000. As 
in the population-level analysis, excluding fledgling pro- 
ductivity during 2001-03 ensured that all fledglings in 
our analysis had at least three years to acquire a breeding 
territory and be detected. We then related the number 
of banded fledglings produced by each adult to the total 
number of their offspring that had successfully recruited 
to the local breeding population by 2003, We used fledg- 
ling production by adults during the study period rather 
than lifetime productivity because our interest was in the 
reliability of individual fledging success as a measure of 
an individual’s fitness potential, regardless of how often 
they bred or how long they had occupied a breeding 
territory. Nonetheless, lifetime fledgling production was 
captured for nearly all goshawks included in our analysis 
because of the duration of our study relative to the num- 
ber of years goshawks successfully laid eggs (x = 2.1 yr, 
min. = 1 yr, max. = 10 yr, 1991-2004; R. Reynolds un- 
publ. data) . We assumed that adults captured or resight- 
ed at a nest were indeed the biological parents of the 
nestlings banded at that nest. In raptors, females are of- 
ten alone before and during egg laying while their mates 
forage, which could result in extra-pair fertilizations by 
conspecific males (Reynolds and Linkhart 1990, Negro 
et al. 1996, Gavin et al. 1998). Although extra-pair fertil- 
izations could confound the relationship between fledg- 
ling success and recruitment, genetic evidence demon- 
strated that extra-pair fertilizations are infrequent for 
goshawks on the Kaibab Plateau (Gavin et al. 1998). 

Data Analysis. At the population level, we used Spear- 


September 2005 


Biology 


213 


Table 1. Northern Goshawk survey effort, banding activity, and local recruitment (banded nestlings from annual 
cohorts that eventually returned to breed) on the Kaibab Plateau, Arizona, 1991-2003. 


Year 

Territories 

Surveyed 

Used 

Nests (%)^ 

Adults 

Banded 

Nestlings 

Banded 

Nestlings 

Returned 

1991 

37 

36 (97) 

48 

46 

8 

1992 

64 

59 (92) 

43 

32 

2 

1993 

82 

67 (82) 

30 

62 

4 

1994 

88 

21 (24) 

13 

18 

2 

1995 

99 

53 (54) 

39 

52 

7 

1996 

105 

46 (44) 

18 

41 

1 

1997 

106 

31 (29) 

15 

36 

9 

1998 

109 

58 (53) 

37 

84 

9 

1999 

113 

57 (50) 

14 

76 

5 

2000 

120 

66 (55) 

33 

111 

1 

2001 

120 

30 (25) 

5 

31 

0 

2002 

121 

21 (17) 

5 

16 

0 

2003 

121 

10 (8) 

2 

9 

— 


® Percent of total territories under study that contained used nests (eggs laid) . 


man correlation coefficients (r^) to characterize the 
strength of the relationship between the size of each an- 
nual fledgling cohort and the number of local recruits 
originating from each cohort. We estimated the expected 
number of fledglings recruited from each annual cohort 
by multiplying our overall estimate of local recruitment 
by cohort size. A chi-square analysis was then used to eval- 
uate possible deviations from the expected number of 
fledglings recruited from each cohort. At the individual 
level, we used generalized linear models (PROG GEN- 
MOD; SAS Institute 1999) to investigate our prediction 
that total fledging production by individual adults would 
be positively related to the number of local recruits each 
produced. Specifically, we used Poisson regression in a 
log-linear model in which a count of recruits produced 
per adult was the response variable and the number of 
banded fledglings produced per adult was assessed as an 
explanatory variable. Chi-square tests or a Fisher’s exact 
test were used to examine potential sex-related differ- 
ences in the age at first breeding and local recruitment. 
All analyses were conducted using SAS (ver. 8.2). Results 
are reported as means ± SE, with 95% confidence inter- 
vals (Cl). 

Results 

The number of territories surveyed increased 
from 37 in 1991 to 121 in 2003 (Table 1). All but 
two of the 121 territories contained nests with eggs 
or young in one or more years of the study. The 
exceptions were territories where goshawks occu- 
pied newly built or reconstructed old nests, but did 
not lay eggs. The percentage of territories under 
study that contained nests with eggs or young var- 
ied substantially among years, ranging from 8% in 
2003 to 97% in 1991 (Table 1). This resulted in 
highly variable fledgling production among years. 


ranging from only 9 fledglings in 2003 to 111 in 
2000. Of the 13-yr total of 555 nests with eggs, 447 
(80.5%) adult pairs fledged 897 young successfully 
(614 of which were banded), and 108 (19.5%) 
pairs lost their clutch during the incubation or 
nestling stages. 

Age at First Breeding and Local Recruitment. In 

the 10 yr that we included for fledgling production 
(1991-2000), we banded 558 nestlings (278 fe- 
males and 280 males) at 494 nests (Table 1). Forty- 
eight (8.6%) of these nestlings returned to breed 
on the study area-26 (9.4%) females and 22 
(7.9%) males. Nestling return rates were similar 
between the sexes (x^ = 0.40, df = 1, P = 0.53). 
Ages at first breeding could be determined with 
confidence for 28 of 48 fledglings that returned to 
breed (17 of 26 females and 11 of 22 males; Fig. 
1). These hawks initiated breeding between 2-8 yr 
of age (3.5 ± 0.32 yr, 95% Cl = 2.84—4.16, median 
= 3 yr) . Only four hawks (all females) were older 
than 4 yr of age at first breeding. Differences be- 
tween sexes in the age at first breeding were not 
supported (Fisher’s Exact Test: P = 0.49). No 
hawks in first-year plumage were observed breed- 
ing at or occupying a nest during the 13-yr study. 
Two females banded as nestlings were later resight- 
ed in adult plumage occupying inactive breeding 
territories on the study area, and two locally band- 
ed nestlings were found breeding in forests beyond 
the Kaibab Plateau (Wiens 2004). 

Population Productivity. We found no evidence 


214 


Wiens and Reynolds 


VoL. 39, No. 3 



Age (years) 


Figure 1. Distribution of ages at first breeding for 28 known-aged Northern Goshawks recaptured on their first 
breeding attempt on the Kaibab Plateau, Arizona, 1991-2003. 


that the number of fledglings recruited from an- 
nual cohorts was correlated with cohort size (r^ = 
0.10, P = 0.79, N = 10; Fig. 2). Likewise, there was 
no evidence of a correlation between annual esti- 
mates of fledging production and recruitment 
when sexes were analyzed separately (females: = 

0.24, P = 0.50; males: r, = 0.11, P = 0.76). The 
number of local recruits produced was significantly 
higher than expected for the 1991 (8 observed, 3.9 
expected; ~ 4.13, df = \, P = 0.04) and 1997 


(9 observed, 3.1 expected; — 11.25, df = 1, P< 
0.01) fledgling cohorts, but much lower than ex- 
pected for the 2000 fledgling cohort (1 observed, 
9.5 expected; x^ ~ 7.65, df = 1, P < 0.01), We 
found no other deviations in numbers of observed 
versus expected recruits in any other fledgling co- 
hort. Thus, the 2000 fledgling cohort was the only 
one of the 10 cohorts in which significantly fewer 
fledglings returned to breed than expected. To en- 
sure that including this year did not unduly bias 



Figure 2. Number of locally produced Northern Goshawk fledglings recruited into the breeding population on the 
Kaibab Plateau, Arizona, relative to the number of young fledged within each annual cohort, 1991-2000. Each point 
represents one year. 


September 2005 


Biology 


215 


(0 

3 

■o 

< 


0 ) 

n 

E 




■ Females 
□ Males 


n 


"I — 


0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 

Number of Fledglings Produced 


Figure 3. Fledgling production of 156 female and 134 male color-marked adult (>2-yr-old) Northern Goshawks 
breeding between 1991 and 2000 on the Kaibab Plateau, Arizona. 


our results, we reanalyzed the data without this co- 
hort. Although this improved the relationship 
slightly, the size of a fledgling cohort was still a 
poor predictor of the total number of goshawks 
that returned to breed from the cohort (sexes 
combined: = 0.46, P = 0.21, N — 9; females: 

= 0.52, P = 0.15; males: r, = 0.46, P = 0.21). 

Individual Productivity. Fledging success was cal- 
culated for 156 females and 134 males breeding 
during 1991—2000 (Table 1). During this period, 
females spent between 1-7 yr as breeders (2.18 ± 
0.11 yr; 95% Cl = 1.96-2.40) and fledged between 
0 and 15 young (3.39 ± 0.23 young; 95% Cl = 
2.93-3.85). Similarly, males also spent between 1 
and 7 years as breeders (1.96 ± 0.11 yr; 95% Cl = 
1.73-2.18) and fledged between 0 and 15 young 
(2.98 ± 0.23 young; 95% Cl = 2.51-3.44). The dis- 
tribution of fledgling production among breeders 
was highly skewed, with most individuals producing 
few or no fledglings and only a few individuals hav- 
ing extremely high success (Fig. 3). In total, 156 
adult females produced 529 fledglings and 134 
adult males produced 399 fledglings (the remain- 
ing 29 and 159 banded nestlings were parented by 
unknown adult females and males, respectively). 
Thirty-six adults (19 females and 17 males) laid 
eggs but failed to produce fledglings. 

The recruitment success of an adult’s offspring 
was significantly related to the total number of 
fledglings the adult produced (sexes combined: re- 
gression slope coefficient, $ = 0.22 ± 0.03, Wald 


= 74.77, df = I, P< 0.01, N = 290; females: p 
= 0.22 ± 0.04, = 38.2, df = 1, P < 0.01, N = 

156; males: 3 = 0.22 ± 0.04, x^ ^ 37.3, df = 1, P 
< 0.01, N = 134; Fig. 4). We had two concerns 
regarding this result. First, we suspected that fledg- 
lings produced during the 2000 breeding season 
might not have had sufficient time to be recruited 
by 2003 (as was our concern in the population- 
level analysis) . Second, there were 33 new unband- 
ed adults (17 females and 16 males) who first bred 
in the study population in 2000 (Table 1). Several 
of these individuals also bred in subsequent years, 
so our initial analysis may have not represented 
their contributions adequately to fledgling produc- 
tion. To address these concerns, we reanalyzed the 
data without individual productivity and recruit- 
ment data from the 2000 breeding season. Exclud- 
ing the 2000 breeding season had little effect on 
the strength of the relationship (sexes combined: 
p = 0.28 ± 0.03, x^ ^ 71.0, df = 1, P < 0.01, N - 
257; females: g = 0.28 ± 0.05, x^ = 37.1, df = 1, 
P < 0.01, N = 139; males: p = 0.28 ± 0.05, x^ = 
36.5, df = 1, P < 0.01, A = 118). Thus, fledging 
success of individual goshawks could be taken as a 
reasonable index of their contributions to the fu- 
ture breeding population. 

Both sexes exhibited similar patterns of variance 
in fledgling production and in the number of re- 
cruits they produced (Figs. 3, 4) . Moreover, fledg- 
ing success and contributions of offspring to the 
future breeding population varied extensively 


216 


Wiens and Reynolds 


VoL. 39, No. 3 



Number of Young Fledged Per Adult 

Figure 4. Number of descendants of individual female {N = 156) and male (N = 134) adult (>2-yr-old) Northern 
Goshawks that successfully recruited into the breeding population in relation to the number of fledglings each adult 
produced between 1991 and 2000 on the Kaibab Plateau, Arizona. Each point represents an individual adult. Trend 
lines determined by Poisson regression with a log-link function (females: y = exp [—2.217 + 0.221x]; males: y = 
exp [-2.056 + 0.217x]). 


among territorial hawks; 20% of adults breeding 
during 1991-2000 contributed nearly 50% of the 
fledglings and 83% of the local recruits produced 
during this time (Fig. 5). Of 290 breeding gos- 
hawks, 73 (25.2%) produced at least one local re- 
cruit each, seven (2.4%) produced two recruits 
each, and three (1%) produced three recruits 
each. One male produced four local recruits, but 
no female produced more than three. In tracking 
lineages up to three generations, three females left 
one breeding descendant each who also produced 
young that were recruited, and two males left one 


breeding descendant each who also produced 
young that were recruited. 

Discussion 

When measured at the population level, our re- 
sults failed to support the hypothesis that fledging 
success is a reliable index of local recruitment in 
goshawks. When measured on an individual basis, 
however, adult fledging success was significantly re- 
lated to local recruitment, indicating that total 
fledgling production by individuals was a more re- 
liable index of reproductive success than a mea- 



Cumulative Proportion of Breeding Adults 

Figure 5. Individual variation in total fledgling production and reproductive success (number of offspring that were 
local recruits) among 290 color-marked adult (>2-yr-old) Northern Goshawks breeding between 1991 and 2000 on 
the Kaibab Plateau, Arizona. Whereas 254 (88%) adults successfully fledged young, only 73 (25%) made contributions 
to the future local breeding population. 


sure of annual productivity. Another important 
finding was that reproductive success varied widely 
among breeding adults. Thus, the relationship be- 
tween fledging success and recruitment appeared 
to be highly dependent upon the disproportionate 
contributions of fledglings and recruits made by a 
relatively small number of adults rather than over- 
all population productivity. Our results further 
demonstrate that recruitment of goshawks on the 
Kaibab Plateau was a gradual process (as shown by 
a wide range among individuals in the age at first 
breeding) and that local recruitment was relatively 
low (8.6%). Given that 25% of adults fail to return 
to reclaim their territory each year (Reynolds et al. 
2004), external recruitment (i.e., immigration) 
would need to be ca. 16% to maintain a stable 
breeding population. 

Age at First Breeding and Local Recruitment. 

Goshawks are capable of breeding in their first year 
of life, and the proportion of breeders in first-year 
plumage has been reported to be as high as 35- 
40% (McGowan 1975, Reynolds and Wight 1978, 
Speiser and Bosakowski 1991, Younk and Bechard 
1994) . The proportion of young hawks in a breed- 
ing population may be reflective of density-depen- 
dent processes driven by the availability of nesting 


territories, food, or mates (McGowan 1975, New- 
ton 1989b, Ferrer et al. 2004). As breeder density 
and survival increases, fewer territories are avail- 
able to prospective breeders, forcing socially sub- 
ordinate younger hawks to wait, perhaps several 
years, to gain a breeding vacancy. Observations of 
individuals breeding in non-adult plumage may, 
therefore, reflect an important buffer mechanism 
compensating for an increased adult mortality rate 
(Ferrer et al. 2004) . On the Kaibab Plateau, several 
demographic features such as a temporally stable 
adult survival rate (Reynolds et al. 2004), a high 
density of breeding territories (8.6/100 km^; Reyn- 
olds et al. 2005), and strong adult fidelity to terri- 
tory and mate (R. Reynolds unpubl. data) suggest 
that young goshawks prospecting for breeding op- 
portunities are faced with the alternatives of delay- 
ing breeding altogether or dispersing to recruit 
elsewhere (Wiens 2004) . That some individuals re- 
main (or return) as floaters in the vicinity of their 
natal population on the Kaibab Plateau has been 
confirmed by radiotelemetry (Wiens 2004) and a 
quick replacement of territorial hawks following 
mortality (Reynolds and Joy in press). Collectively, 
these features could explain the relatively ad- 
vanced age of first-time breeders as well as the low 


218 


Wiens and Reynolds 


VoL. 39, No. 3 


local recruitment level during our study. It is im- 
portant to note, however, that these characteristics 
(territory density, adult fidelity, breeding age) ap- 
pear to occur at higher levels on the Kaibab Pla- 
teau than reported for goshawks elsewhere. 

Population Productivity. We found that the gos- 
hawk population on the Kaibab Plateau was no 
more successful in producing local recruits in years 
of high fledgling production than in years of low 
fledgling production. Thus, contrary to predic- 
tions, the size of an annual fledgling cohort was a 
poor predictor of local recruitment. In contrast, 
Weatherhead and Dufour (2000) found that local 
recruitment increased disproportionately with the 
size of Red-winged Blackbird fledgling cohorts, 
even though the overall return rate of banded nest- 
lings in that study was small (2.4%) relative to our 
estimate for goshawks (8.6%). The strength of the 
relationship between population productivity and 
local recruitment, and our inability to detect such 
a relationship, may largely depend on the spatial 
extent of the study area relative to the extent of 
natal dispersal. For example, Red-winged Black- 
birds disperse up to 40 km from natal territories 
(Moore and Dolbeer 1989), whereas young gos- 
hawks disperse up to 440 km from the Kaibab Pla- 
teau (Wiens 2004). The lack of a population-level 
relationship between fledging success and recruit- 
ment in our study may, therefore, simply reflect the 
difference between observing local recruitment 
within a relatively small subpopulation relative to 
general recruitment taking place over a regionally 
fragmented population that is connected by natal 
dispersal. Several lines of evidence indicate that 
most juveniles disperse long distances beyond the 
Kaibab Plateau in their first fall (Wiens 2004) . Un- 
fortunately, with the exception of two known cases 
of juvenile emigration, no information on the ex- 
ternal recruitment success of locally-produced 
hawks exists for this study population. 

Aside from a potentially inappropriate scale of 
investigation relative to the extent of the goshawk 
recruitment process, our data may have failed to 
support our hypothesis at the population level be- 
cause of a lack of breeding opportunities during 
the later years of the study. During 2001-03, ex- 
treme drought conditions in northern AZ likely led 
to significant declines in goshawk prey populations 
(Salafsky 2004), and few adult pairs on the Kaibab 
Plateau attempted to breed. Under such poor 
breeding conditions, opportunities for inexperi- 
enced hawks attempting to breed for their first 


time were reduced. For the most part, adult gos- 
hawks could only be captured or resighted when 
breeding, so newly recruiting hawks that acquired 
a breeding territory but did not lay eggs may have 
gone undetected during 2001-03. Poor breeding 
conditions during the later years of our study could 
also explain the lower than expected recruitment 
success of fledglings produced in the 2000 breed- 
ing season. However, removing the 2000 cohort 
from the analysis failed to substantially improve the 
relationship between population productivity and 
local recruitment. Our chi-square analysis showed 
that fledglings produced in 1991 and 1997 were 
twice as likely to be recruited as expected. With the 
exception of the 2000 fledgling cohort, we found 
no other significant deviations in the number of 
local recruits expected. We suggest that high juve- 
nile survival, density-dependent effects, or changes 
in resource availability may have contributed to the 
disproportionately high recruitment success of 
fledglings produced during 1991 and 1997. 

Individual Productivity. If fitness is defined as 
the contribution of an individual’s genotype to sub- 
sequent generations proportional to that of other 
individuals (Lincoln et al. 1998), then our results 
clearly show a difference in fitness among gos- 
hawks on the Kaibab Plateau. Our results indicated 
that the likelihood of an adult contributing off- 
spring to the future breeding population on the 
Kaibab Plateau increased disproportionately with 
the number of fledglings it produced. Therefore, 
goshawks appear to be similar to other monoga- 
mous raptor species in that reproductive success 
varies widely among individuals (Newton 1989b, 
Hakkarainen et al. 1997, Marti 1997). As would be 
expected for a monogamous raptor in a popula- 
tion showing high fidelity to territory and mate 
and equal survival rates of males and females, the 
sexes showed similar patterns of variance in repro- 
ductive success. Indeed, a few territorial pairs ex- 
hibited disproportionately high fitness levels in 
terms of the number of years they bred, total fledg- 
ling production, and the number of their offspring 
that survived to become breeding adults. Factors 
not explored in this study, such as breeding age, 
habitat quality, individual quality (as determined 
by genetic forces), or climate may contribute to 
individual heterogeneity in goshawk reproductive 
success. 

On the Kaibab Plateau, the mean reproductive 
lifespan of adults (first breeding to disappearance) 
is 2 yr, but a few adults were found to breed for as 


September 2005 


Biology 


219 


many as 10 yr (R. Reynolds unpubl. data). Adults 
who breed more often are likely to be more ex- 
perienced, produce more fledglings, and show the 
highest fitness potential in terms of survival and 
reproduction (Gam et al. 1998). Newton and Roth- 
ery (2002) , for example, demonstrated that almost 
all aspects of breeding performance in female Eur- 
asian Sparrowhawks improved with age, but that 
the degree of improvement lessened with each suc- 
cessive year of life. In Denmark, Nielsen and 
Drachmann (2003) showed that fledgling produc- 
tion of European goshawks increased with female 
age from 1-7 yr, but declined thereafter. Similarly, 
Reynolds et al. (1994) reported a difference in 
fledging production between young-adult (second- 
year) and full-adult goshawks. These authors attri- 
buted changes in reproductive success to age-relat- 
ed trends in foraging efficiency, which could be 
buffered by pairing with more experienced mates. 

Variation among individuals in fitness levels can 
have substantial effects on population growth, sta- 
bility, and persistence (Bj0rnstad and Hansen 
1994, Conner and White 1999). Individual varia- 
tion within raptor populations may be generated 
by temporal or spatial variations in resource avail- 
ability and habitat quality (Hakkarainen et al. 1997, 
Franklin et al. 2000). Of particular interest is the 
relationship between individual fitness and habitat 
quality, as it is commonly supposed that birds pre- 
fer the habitat that will confer the greatest fitness 
(Fretwell and Lucas 1970). For goshawks, hetero- 
geneity in fitness levels among territorial adults 
could be caused by spatial or temporal variations 
in food abundance, resource availability (as mea- 
sured by qualitative differences in forest composi- 
tion and structure among breeding areas) , and in- 
dividual hawk quality. However, recent efforts to 
detect spatial heterogeneity in goshawk reproduc- 
tive parameters have had limited success. On the 
Kaibab Plateau, Joy (2002) ranked 101 goshawk 
territories as “high” or “low” quality based on dif- 
ferences in egg laying frequency and fledgling pro- 
duction. The length of time an adult female re- 
mained on a territory, breeder age, and the 
amount or spatial configuration of vegetative types 
within territories explained little of the variation 
(Joy 2002). Elsewhere, McClaren et al. (2002) 
found minimal spatial variation in the number of 
young fledged per nest among goshawk nest areas 
in three study sites in western North America. Giv- 
en that these studies focused on territories (Joy 
2002) and nest areas (McClaren et al. 2002) rather 


than individual goshawks, we suggest that unde- 
tected turnovers between territory or nest occu- 
pants of different age, experience, or genetic qual- 
ity could mask existing spatial patterns in 
reproduction. For the purposes of identifying the 
determinants of habitat quality for goshawks, we 
believe our finding of individual variation in repro- 
ductive success highlights the need to partition the 
effects of individual hawk quality from habitat qual- 
ity. By controlling for individual hawk quality, re- 
searchers can more precisely estimate the effects 
of habitat versus non-habitat factors on compo- 
nents of goshawk fitness. Given the hypothesized 
role of forest composition and structure in gener- 
ating individual variation in goshawk reproductive 
performance, long-term demographic research 
based on color-marked individuals can provide a 
powerful tool to guide goshawk conservation and 
management. 

Acknowledgments 

Our research was supported by the USDA Forest Ser- 
vice (Southwest Region), Rocky Mountain Research Sta- 
tion, the North Kaibab Ranger District, and a grant from 
the Heritage Program, AZ Game and Fish Department. 
We are grateful to the many field technicians and vol- 
unteers who helped collect data on Northern Goshawks 
during 1991-2003. The North Kaibab Ranger District 
graciously provided housing and logistical support dur- 
ing this study. Special thanks to S.M. Joy, C.M. Erickson, 
J.S. Lambert, J.C. Seyfried, S.R. Salafsky, and S. Bayard de 
Volo for their help with data entry and summarization. 
R.M. King provided statistical advice. We thank B.R. 
Noon, P.J. Weatherhead, J.A. Smallwood, C. Crocker-Bed- 
ford, and C.W. Boal for providing constructive comments 
which improved this manuscript. 

Literature Cited 

Bayard de Volo, S., R.T. Reynolds, J.R. Topinka, B. May, 
AND M.F. Antolin. 2005. Population genetics and ge- 
notyping for mark-recapture studies of Northern Gos- 
hawks {Accipiter gentilis) on the Kaibab plateau, Ari- 
zona. J. Raptor Res. 39:286-295. 

Bjornstad, O.N. and T.F. Hansen. 1994. Individual vari- 
ation and population dynamics. Oikos 69:167-171. 
Byholm, P., P. Saurola, H. Linden, and M. Wikman. 
2003. Causes of dispersal in Northern Goshawks {Ac- 
dpiter gentilis) in Finland. Auk 120:706-716. 

Cam, E., J.E. Hines, J. Monnat, J.D. Nichols, and E. Dan- 
chin. 1998. Are adult nonbreeders prudent parents? 
The Kittiwake model. Ecology 79:2917—2930. 

Conner, M.M. and G.C. White. 1999. Effects of individ- 
ual heterogeneity in estimating the persistence of 
small populations. Nat. Res. Model. 12:109—127. 
Detrich, P.J. and B. Woodbridge. 1994. Territory fidel- 
ity, mate fidelity, and movements of color-marked 


220 


Wiens and Reynolds 


VoL. 39, No. 3 


Northern Goshawks in the southern Cascades of Cal- 
ifornia. Stud. Avian Biol, 16:130-132. 

Endler, J.A. 1986. Natural selection in the wild. Prince- 
ton Univ. Press, Princeton NJ U.S.A. 

Ferrer, M., F. Otalora, and J.M. Garcia-Ruiz. 2004. 
Density-dependent age of first reproduction as a buff- 
er affecting persistence of small populations. Ecol. 
Appl. 14:616-624. 

Franklin, A.B., D.R. Anderson, R.J. Gutierrez, and K.P. 
Burnham. 2000. Climate, habitat quality, and fitness 
in Northern Spotted Owl populations in northwest- 
ern California. Ecol. Monographs 70:539—590. 

Fretwell, S.D. and H.L. Lucas, Jr. 1970. On territorial 
behavior and other factors influencing habitat distri- 
bution in birds. I. Theoretical development. Acta Bio. 
19:16-36. 

Gavin, T.A., R.T. Reynolds, S.M. Joy, D.G. Leslie, and B. 
May. 1998. Genetic evidence for low frequency of ex- 
tra-pair fertilizations in Northern Goshawks. Condor 
100:556-560. 

Greenwood, P.J. and P.H. Harvey. 1982. The natal and 
breeding dispersal in birds. Annu. Rev. Ecol. Syst. 13: 
1 - 21 . 

Hakkarainen, H., V. KorvxJVEN, and E Korpimaki. 1997. 
Reproductive success and parental effort of Teng- 
raalm’s Owls: effects of spatial and temporal variation 
in habitat quality. Ecoscience 4:35-42- 

Joy, S.M. 2002. Northern Goshawk habitat on the Kaibab 
National Forest in Arizona: factors affecting nest lo- 
cations and territory quality. Ph.D. dissertation, Col- 
orado State Univ., Fort Collins, CO U.S.A. 

Keedwell, RJ. 2003. Does fledging equal success? Post- 
fledging mortality in the Black-fronted Tern. J. Field 
Ornithol. 74:217-221. 

Kennedy, P.L. 1997. The Northern Goshawk {Accipiter gen- 
tilis atricapillus) : is there evidence of a population de- 
cline? J. Raptor Res. 31:95-106. 

AND J.M. Ward. 2003. Effects of experimental 

food supplementation on movements of juvenile 
Northern Goshawks {Accipiter gentilis atricapillus) . Oe- 
cologia 134:284—291. 

Lincoln, R., G. Boxshall, and P. Clark. 1998. A dictio- 
nary of ecology, evolution, and systematics, 2nd Ed. 
Cambridge Univ. Press, Cambridge, U.K. 

Marti, C.D. 1997. Lifetime reproductive success of Barn 
Owls near the limit of the species’ range. Auk 114: 
581-592. 

Martin, K., P.B. Stacey, and C.E. Braun. 2000. Recruit- 
ment, dispersal, and demographic rescue in spatially- 
structured White-tailed Ptarmigan populations. Con- 
dor 102:503-516. 

McClaren, E.L., P.L. Kennedy, and S.R. Dewey 2002. Do 
some Northern Goshawk nest areas consistently 
fledge more young than others? Condor 104:343-352. 

McGowan, J.D. 1975. Distribution, density, and produc- 
tivity of goshawks in interior Alaska. Federal Aid Wild- 
life Restoration Project. Rep. W-17-4, W-17-5, W-17- 


6, Job 10.6A. Alaska Department of Fish and Game. 
Anchorage, AK. 

Moore, W.S. and R.A. Dolbeer. 1989. The use of band- 
ing recovery data to estimate dispersal rates and gene 
flow in avian species: case studies in the Red-winged 
Blackbird and Common Grackle. Condor 91:242-253. 

Negro. JJ-. M. Villarroel, J.L. Tella, U. Kuhnlein, F. 
Hiralso, J.A. DonAzar, and D.M. Bird. 1996. DNA 
fingerprinting reveals low incidence of extra-pair fer- 
tilizations in the Lesser Kestrel. Anim. Behav. 51:935— 
943. 

Newton, 1. 1989a. Sparrowhawk. Pages 279-296 in I. 
Newton [Ed..], Lifetime reproduction in birds. Aca- 
demic Press, New York, NY U.S.A. 

(Ed.). 1989b. Lifetime reproduction in birds. Ac- 
ademic Press, New York, NY U.S.A. 

AND P. Rothery. 2002. Age-related trends in dif- 
ferent aspects of the breeding performance of indi- 
vidual female Eurasian Sparrowhawks {Accipiter nisus) 
Auk 119:735-748. 

Nielsen, J.T. and J. Drachmann. 2003. Age-dependent 
reproductive performance in Northern Goshawks 
{Accipiter gentilis) . Ibis 145:1-8. 

Postupalsky, S. 1989. Osprey. Pages 297-313 in I. New- 
ton [Ed.], Lifetime reproduction in birds. Academic 
Press, New York, NY U.S.A. 

Rasmussen, D. I. 1941. Biotic communities of the Kaibab 
Plateau, Arizona. Ecol. Monographs 11:230—273. 

Reich, R.M, S.M. Joy, and R.T. Reynolds. 2004. Predict- 
ing the location of Northern Goshawk nests: model- 
ing the spatial dependency between nest locations 
and forest structure. Ecol. Modeling 176:109—133. 

Reynolds, R.T. and S.M. Joy. In press. Demography of 
Northern Goshawks in Northern Arizona, 1991-96. 
Stud. Avian Biol. 

, , AND D.G. Leslie. 1994. Nest productivity, 

fidelity, and spacing of Northern Goshawks in north- 
ern Arizona. Stud. Avian Biol. 16:106-113. 

AND B.D. Linkhart. 1990. Extra-pair copulation 

and extra-range movements in Flammulated Owls. Or- 
nis. Scand. 21:74—77. 

, G.C. White, S.M. Joy, and R.W. Mannan. 2004. 

Effects of radiotransmitters on Northern Goshawks, 
do tailmounts lower survival of breeding males? J. 
Wildl. Manag. 68:25-32. 

, J.D. Wiens, S.M. Joy, and S.R. Salafsky 2005. 

Sampling considerations for demographic and habitat 
studies of Northern Goshawks. J. Raptor Res. 39:274- 
285. 

AND H.M. Wight. 1978. Distribution, density, and 

productivity of Accipiter hawks breeding in Oregon. 
Wilson Bull. 90:182-196. 

Salafsky, S.R. 2004. Covariation between prey abundance 
and Northern Goshawk fecundity on the Kaibab Pla- 
teau, Arizona. M.S. thesis, Colorado State Univ., Fort 
Collins, CO. U.S.A. 

SAS Institute, 1999. SAS/STAT User’s guide, version 8, 


September 2005 


Biology 


221 


Volumes 1, 2, and 3. SAS Institute, Inc., Cary, NC 
U.S.A. 

Saurola, P. 1989. Ural Owl. Pages 327-345 in I. Newton 
[Ed.], Lifetime reproduction in birds. Academic 
Press, New York, NY U.S.A. 

SONSTHAGEN, S.A., S.L. TALBOT, AND C.M. WHITE. 2004, 
Gene flow and genetic characterization of Northern 
Goshawks breeding in Utah. Condor 106:826-836. 

Speiser, R. and T. Bosakowski. 1991. Nesting phenology, 
site fidelity, and defense behavior of Northern Gos- 
hawks in New York and New Jersey. J, Raptor Res. 25: 
132-135. 

Stacey, P.B. and M. Taper. 1992. Environmental variation 
and the persistence of small populations. Ecol. Appl. 
2:18-29. 

Squires, J.G. and R.T. Reynolds. 1997. Northern Gos- 


hawk {Accipiter gentilis) . In A. Poole and F. Gill [Eds.], 
The birds of North America, No. 298. The Birds of 
North America, Inc., Washington, DC U.S.A. 

Weatherhead, PJ AND K.W. Dufour. 2000. Fledging suc- 
cess as an index of recruitment in Red-winged Black- 
birds. Auk 117:627-633. 

Wiens, J.D. 2004. Post-fledging survival and natal dis- 
persal of Northern Goshawks in Arizona. M.S. thesis, 
Colorado State Univ., Fort Collins, CO U.S.A. 

Younk, J.V. and MJ. Bechard. 1994. Effect of gold min- 
ing activity on Northern Goshawks breeding in Ne- 
vada’s Independence and Bull Run mountains. Rap- 
tor Research Center, Department of Biology, Boise 
State University, Boise, ID U.S.A. 

Received 18 February 2004; accepted 5 April 2005 

Associate Editor: Clint Boal 


J Raptor Res. 39 (3) :222-228 
© 2005 The Raptor Research Foundation, Inc. 

PRODUCTIVITY AND MORTALITY OF NORTHERN GOSHAWKS 

IN MINNESOTA 

Clint W. Boal^ 

U.S. Geological Survey, Texas Cooperative Fish and Wildlife Research Unit, Texas Tech University, 

Lubbock, TX 79409 US. A. 

David E. Andersen 

U.S. Geological Survey, Minnesota Cooperative Fish and Wildlife Research Unit, University of Minnesota, 

St. Paul, MN 53108 U.S. A. 

Patricia L. Kennedy 

Eastern Oregon Agricultural Research Station and Department of Fisheries and Wildlife, Oregon State University, 

P.O. BoxE, Union, OR 97883 U.S. A. 

Abstract. — Compared to other regions of North America, little information exists regarding Northern 
Goshawk {Accipiter gentilis) ecology and population dynamics in the western Great Lakes Region. We 
examined productivity and nesting habitat characteristics of goshawks in Minnesota from 1998-2001. 
Apparent nesting success varied annually from as low as 38% to as high as 83%. The Mayfield estimate 
of daily survival for nests was 0.992 ± 0.002 (SE). The mean fledgling number across years was 1.85 ± 
0.14 for successful nests and 1.14 ± O.D for all nesting attempts. Twenty-one percent of all nesting 
attempts failed, primarily due to predation or suspected predation (52%) and inclement weather (35%). 
Overall, productivity of goshawks in Minnesota was at the lower end of the range reported in other 
studies across western North America, which is not atypical for peripheral populations. During the 3-yr 
study, we recorded mortalities of nine (four males and five females; eight radio-marked and one un- 
marked) adult goshawks — causes of mortality were avian (33%) and mammalian (22%) predation, hu- 
man persecution (22%), and unknown causes (22%). Fifty-six percent of mortalities occurred during 
the breeding season, and 44% occurred during the winter. Based on radiotelemetry data, we estimated 
adult annual survival to be 74 ± 7.8%, which is similar to survival estimated using mark-recapture 
analysis in three western North America studies. 

Key Words: Northern Goshawk, Accipiter gentilis; breeding, Minnesota; mortality; productivity. 


PRODUCTIVIDAD YMORTALIDAD DE ACCIPITER GENTILIS YM MINNESOTA 

Resumen. — Existe poca informacion sobre la ecologia y la dinamica poblacional de Accipiter gentilis en el 
oeste de la region de Grandes Lagos comparado con otras regiones de America del Norte. Examinamos 
la productividad y las caracteristicas del ambiente de nidificacion de A. gentilis en Minnesota desde 1998 
hasta 2001. El exito de nidificacion aparente vario anualmente de modo drastico, desde 38% a 83%. El 
estimado de Mayfield de la supervivencia diaria de los nidos fue 0.992 ± 0.002 (EE). El numero medio 
de volan tones a lo largo de los afios fue 1.85 ± 0.14 para los nidos exitosos y 1.14 ± 0.17 para todos los 
intentos de nidificacion. El 21% de todos los intentos de nidificacion fracaso, debido principalmente a la 
depredacion o a la supuesta depredacion (52%) y a las inclemencias del clima (35%). En total, la pro- 
ductividad de A. gentilis en Minnesota estuvo en el extremo inferior del rango reportado en otros estudios 
para el oeste de America del Norte, lo cual no es atfpico para poblaciones perifericas. Durante los tres 
anos de estudio, registramos la muerte de nueve adultos de A. gentilis (4 machos y 5 hembras; 8 marcados 
con transmisores y 1 sin marcar). Las causas de la mortalidad fueron depredacion por aves (33%) y 
mamiferos (22%), persecucion humana (22%) y causas desconocidas (22%). El 56% de las muertes ocu- 
rrieron durante la estacion reproductiva y el 44% durante el invierno. Basados en datos de radio telemetria, 
estimamos que la supervivencia anual de los adultos fue del 74 ± 7.8%, lo cual es similar a la supervi- 
vencia estimada usando analisis de captura-recaptura en tres estudios del oeste de America del Norte. 

[Traduccion del equipo editorial] 


^ Email address: cboal@ttu.edu 


222 


September 2005 


Biology 


223 


The Northern Goshawk {Accipiter gentilis) is a 
large, forest-dwelling raptor generally associated 
with mature deciduous, coniferous, or mixed for- 
ests. Possible conflicts between timber harvest 
practices and goshawk habitat requirements have 
led to concern for the species’ status (Kennedy 
1997, United States Fish and Wildlife Service 
1998). The goshawk has been proposed for listing 
several times under the U.S. Endangered Species 
Act and its status has been (and still is) the object 
of considerable litigation. In the western Great 
Lakes Region (WGLR) of North America, the gos- 
hawk is currently listed as a migratory non-game 
bird of management concern by the U.S. Fish and 
Wildlife Service (Region 3) and as a sensitive spe- 
cies by the U.S. Forest Service (Region 9). Few 
studies have examined goshawk productivity (Erd- 
man et al. 1998) and mortality in the WGLR. Re- 
gion-specific information on productivity and mor- 
tality factors is essential for development of sound 
management guidelines, but active management of 
the species in the WGLR has been hampered by 
the lack of data. In 1998, we initiated a broad-based 
ecological study of goshawks in Minnesota (Boal et 
al. 2001, Boal et al. 2003). Herein, we present the 
productivity and mortality data we collected on 
breeding goshawks in Minnesota, 1998-2000. 

Study Area 

The study area encompassed most of northern Min- 
nesota within the Laurentian Mixed-Forest Province 
(Minnesota Department of Natural Resources 2004; Fig. 
1). Goshawks were distributed across the study area (Fig. 
1), but a majority of goshawk nests were located on or 
near the Chippewa National Forest (47°23'N, 94°35'W). 
Study area elevation was ca. 200-400 m. Historical mean 
summer and winter temperatures were 18°C and — 11“C, 
respectively, with maximum and minimum temperature 
records of 40°C and — 46°C, respectively. Vegetation com- 
munities are described in Boal et al. (2003). 

Methods 

Study Population. We did not systematically survey for 
breeding goshawks, so known breeding pairs in a single 
year were likely a relatively small proportion of all gos- 
hawks breeding in the study area (Daw et al. 1998). How- 
ever, the goshawks monitored in this study were all 
known nesting goshawks in Minnesota during the study 
period of 1998-2000 (Boal et al. 2001, 2003). Nests in 
this study were from across the Laurentian Mixed-Forest 
Province of northern Minnesota and were likely repre- 
sentative of the Minnesota landscapes that goshawks use 
for nesting (Boal et al. 2001), but because our sample 
was not randomly selected, our inferences are limited to 
our sample. 

Before this study, few goshawk nesting areas were 
known in Minnesota. We searched known goshawk nest 


stands and areas where goshawks had been seen during 
previous breeding seasons. If a previous year’s nest was 
not occupied, we conducted tree-by-tree searches of the 
stand, up to 500 m from the old nest (if the stand was 
sufficiently large for this search pattern) . We also located 
new goshawk nest stands by searching likely areas or fol- 
lowing up on reports of probable goshawk nests located 
serendipitously by personnel from cooperating agencies 
and the timber industry. We considered an area to be 
occupied if one goshawk was observed in or near a 
known nest stand, radio-tagged hawks were located in the 
area, or other evidence of activity was observed (e.g., re- 
cent construction of a nest) . If an area was occupied by 
goshawks, we attempted to locate an occupied nest. An 
occupied nest was defined as a nest with eggs or young 
or the presence of an incubating goshawk. 

Productivity. Once an occupied nest was located, we 
made periodic visits to monitor reproductive success. We 
considered goshawks to be nesting if a female was ob- 
served in an incubation position on the nest or during 
later stages of the nesting period when young were ob- 
served in the nest. We considered nestlings to have sur- 
vived to fledge if they attained at least 80% of their first 
flight age (32 d old for goshawks; Boal 1994). We consid- 
ered a nesting attempt as successful if at least one young 
fledged. We estimated both apparent nest success (e.g., 
the proportion of monitored nests known to have 
fledged young) and nest success using the Mayfield esti- 
mate based on exposure days (Bart and Robson 1982). 
Because confidence intervals can be more informative 
than tests of statistical significance (Johnson 1999), we 
assessed differences in productivity by examining overlap 
of 95% confidence intervals. 

Nesting Failure. We attempted to determine cause of 
all nesting failures. In instances where dead adults or 
their remains were found at nests, we conducted in-field 
examinations of each carcass and location of death to 
attempt to identify the cause of death and, if depredated, 
the predator species. For example, claw marks ascending 
the nest tree and teeth marks on the carcass, feathers, 
and radio harness material of radio-tagged birds were in- 
dicative of mammalian predation (Einarsen 1956). In 
contrast, crimping plucks of feathers, stripped bones 
without tooth marks or evidence of mastication, single 
bill bite nips, and scrapes in bones indicated avian pred- 
ators (Einarsen 1956). 

Adult Mortality. There is little information on causes 
of adult mortality for raptors in general and goshawks m 
particular (Squires and Reynolds 1997). In addition to 
assessing causes of mortality of adult goshawks at nest 
sites, our sample of 32 radio-tagged goshawks (Boal et al. 
2003) provided us with an opportunity to examine causes 
and timing of mortality among goshawks in Minnesota. 
We used telemetry to relocate all radio-tagged goshawks 
that died during the course of this study (Boal et al. 
2001). We estimated the annual survival rate with the 
Kaplan-Meier survival model (Kaplan and Meier 1958) as 
modified by Pollock et al. (1989). 

Results 

We located 13, 19, and 21 areas occupied by gos- 
hawks in 1998, 1999, and 2000, respectively. Two 


224 


Boal et al. 


VoL. 39, No. 3 



Minnesota 


Ecortglon Section Nam« 

K Minn«sota and Ontario Paatlanda 
K MInnaaota Df+tt and Lakt Plaint 
Northtm Su pari or Uplandt 
Southarn Suparlor Uplands 
Waatarn Suparlor Uplar>ds 


TO 


140 


210 


280 


S50 Km 


Figure 1. Study area and distribution of Northern Goshawk nests sites (open circles) included in this study, Min- 
nesota 1998-2000. Ecoregions are based on Minnesota Department of Natural Resources (2004), 


additional areas were located by cooperators in 
1998, but were not reported to us until 1999. Al- 
though one of these nesting attempts was verified 
as successful, the two nests were not monitored to 
assess productivity. Thus, we only include the orig- 
inal 13 of 15 areas from 1998 for productivity as- 
sessment. Of the 15 breeding areas occupied in 

1998, 11 (73%) were occupied in 1999. Of 23 
known breeding areas occupied in 1998 and/or 

1999, 13 (57%) were occupied in 2000. Although 
breeding did not occur in all occupied areas, 15 
occupied areas were located in 1998, seven addi- 
tional areas in 1999, and nine additional areas in 

2000, for a total of 31 areas occupied by goshawks 
at least 1 yr during the 3-yr study period. We did 
not monitor productivity at one nest, and two oth- 
ers failed in 1998. Sixteen (84%) pairs of goshawks 
from 19 occupied areas nested in 1999, and 15 
(71%) pairs from 21 occupied areas nested in 
2000. We observed that some areas were occupied 
by non-breeding goshawks. For example, a wid- 
owed female, radio-tagged in 1998, was tracked in 
her breeding area, but did not breed in 1999. Like- 
wise, in 2000 a widowed, non-breeding female 


roamed more widely than she had while breeding 
in 1999, but still occupied her 1999 breeding area. 
A pair that had been radio-tagged and successfully 
nested in 1999 occupied their breeding area, but 
did not nest in 2000. In contrast, after her mate 
died during the winter, one widowed female 
moved 15 km to pair with a male in a previously 
unknown breeding area the following spring. 

Productivity. We assessed success at 43 and pro- 
ductivity at 42 goshawk nests. Nesting success var- 
ied considerably among years, with a high of 83% 
in 1998 and a low of 37% in 1999. We observed 
that 67% of nesting attempts fledged young suc- 
cessfully in 2000 and the 3-yr mean for fledging 
success was 62 ± 23.4% (SE). Mayfield estimates 
for daily survival were 0.9998 ± 0.0006 (SE) in 
1998, 0.985 ± 0.005 in 1999, and 0.993 ± 0.005 in 
2000, with an overall daily survival rate of 0.992 ± 
0.002. Based on a 32-d incubation period (Squires 
and Reynolds 1997) and a 32-d period to 80% of 
first flight age (Boal 1994), Mayfield estimates of 
nest success were 99% in 1998, 39% in 1999, 65% 
in 2000, and 59% over the 3-yr study period. 

Goshawk nests fledged a mean of 1.75 ± 1.05 



September 2005 


Biology 


225 


young in 1998 (N ~ 12), 0.81 ± 1.17 young in 1999 
(N= 16), and 0.93 ± 0.80 {N— 15) young in 2000. 
Fledglings per nesting attempt were not statistically 
different between 1998 and 1999 {x difference = 
0.938; 95% Cl = 0.057-1.820), between 1998 and 
2000 {x difference = 0.817; 95% Cl = 0.082- 
1.550), or between 1999 and 2000 (x difference = 
—0.121; 95% Cl = —0.861—0.619). In contrast, 
when examining only those nests that were suc- 
cessful, goshawks fledged a mean of 2.10 ± 0.74 
young in 1998 {N — 10), 2.17 ± 0.75 young in 1999 
{N = 6), and 1.40 ± 0.52 young in 2000 (A^ =10). 
Fledgling numbers at successful nests were not sta- 
tistically different between 1998 and 1999 (x dif- 
ference = —0.067; 95% Cl = —0.890-0.757), but 
were higher in 1998 (x difference = 0.700; 95% Cl 
= 0.102-1.300) and 1999 (x difference = 0.767; 
95% Cl = 0.895—1.440) than in 2000. Mean num- 
ber of fledglings per nest for all years combined 
was 1.14 ± 1.07 for all nesting attempts and 1.85 
± 0.73 for successful nests only. 

Nesting Failure. Of the 43 goshawk nests moni- 
tored, two failed in 1998, 10 failed in 1999, and 
five failed in 2000. Of these 17 failures, 23% were 
due to mammalian predation, 18% were due to 
avian predation, and we suspected another 12% 
were due to predation, but we were unable to de- 
termine whether the predator was avian or mam- 
malian. Two of the mammalian predations resulted 
in mortalities of adult female goshawks (detailed 
below) . Weather contributed to 35% of nesting fail- 
ures, the majority of which occurred during the 
incubation stage of the nesting period in 1999 
when the region experienced a 10-11 d period of 
almost constant rain. The cause of 12% of nesting 
failures was undetermined. 

Adult Mortality. Nine goshawks, eight of which 
were radio-tagged, died during this study. Five 
(56%; four females and one male) of these nine 
mortalities occurred during the breeding seasons. 
One female and one male were preyed upon by 
Great Horned Owls {Bubo virginianus) , two females 
were killed by mammals, and one female was con- 
sumed by a Red-tailed Hawk {Buteo jamaicensis) . 
The remaining four (44%) mortalities (one female 
and three males) occurred during the winter 
months. The female that died during the winter 
had been shot. The mortality of one male ap- 
peared to also be due to human actions; only the 
radio that had been attached to the male was re- 
covered and it had been obviously cut from the 
body of the goshawk. Furthermore, the radio 


lacked any mastication or pecking marks typical of 
those on goshawks that were depredated. The caus- 
es of death could not be verified for the other two 
male goshawks. 

We excluded the single, non-radio-marked fe- 
male that was killed in 1999 from estimates of sur- 
vival. The Kaplan-Meier estimate for annual surviv- 
al based on 32 radio-tagged goshawks was 74 ± 
7.8%. Because our study was not originally de- 
signed to estimate adult survival, our survival esti- 
mate should be interpreted in the context of our 
sample of 32 radio-marked goshawks captured at 
breeding areas over the 3-yr study period. However, 
these data do provide some insight in the annual 
survival of goshawks in the study area. 

Discussion 

Productivity. We observed annual variability in 
fledglings produced per nest attempt (range = 
0.87—1.85) and per successful nest (range = 1.40— 
2. 1 7) during the 3 yr of this study. Such variability 
is typical of temporal patterns in reproductive suc- 
cess in goshawks (e.g., DeStefano et al. 1994, Ken- 
nedy 1997, McClaren et al. 2002). McClaren et al. 
(2002) found high temporal variability in produc- 
tivity among goshawk nests monitored 4—10 yr in 
three different populations in western North 
America. Within the WGLR, Erdman et al. (1998) 
reported fledgling numbers from Wisconsin gos- 
hawk nests over a 24-yr period, and found a mean 
of 1.7 fledglings per nesting attempt {N = 184) 
and 2.2 per successful nest {N = 138). However, 
Erdman et al. (1998) also indicated that fledging 
rates among nesting attempts decreased from the 
earlier years (1971-81) to later years (1982-92) of 
their study. We do not have historical data for this 
study area to evaluate temporal trends in produc- 
tivity, but the fledging success over the 3-yr period 
of monitoring did not vary statistically. Productivity 
in the Upper Peninsula of Michigan was similar to 
ours, with a reported 1.1 and 1.7 fledglings per 
occupied {N = 36) and successful {N — 24) nests, 
respectively (Lapinski 2000) . Fledgling rate among 
successful nests in our study and others conducted 
in the WGLR appear to be slightly lower than av- 
erage, but well within the range reported in studies 
from the western United States (e.g., Kennedy 
1997, Boal and Mannan 1994, Bull and Hohmann 
1994, Reynolds et al. 1994). 

Nesting Failure. The most common nest preda- 
tor of goshawk nests in North America appears to 
be Great Horned Owls (Kennedy 2003) , but wol- 


226 


Boal et al. 


VoL. 39, No. 3 


verines (Gulo gulo; Doyle 1995) and fishers {Martes 
pennanti; Erdman et al. 1998) are known to prey 
upon goshawk nestlings, and raccoons {Procyon lo- 
tor) are also likely nest predators. Erdman et al. 
(1998) attributed predation by fishers as the pri- 
mary cause of nesting failure among goshawks in 
Wisconsin, but did not provide details for the basis 
of their conclusion, nor did they report the actual 
number of nesting failures due to fishers. Mammal 
depredation (suspected to be fishers) of nests in 
our study was comparatively low (9%), but collec- 
tive depredation (mammalian, avian, unknown) 
caused the failure of 21% of goshawk nests in Min- 
nesota. 

Weather can also influence productivity of gos- 
hawks. Cold weather and rain can reduce the num- 
ber of goshawk pairs attempting to nest (Kostrzewa 
and Kostrzewa 1990) and can lead to egg and chick 
(Zachel 1985) mortality. In our study, inclement 
weather accounted for failure of 12% of all nesting 
attempts. These failures occurred primarily during 
the incubation stage in 1999 when our study area 
experienced a 10-11 d period of almost constant 
rainfall. We suspect that some male goshawks may 
have been unable to provision their mates ade- 
quately during this period, eventually leading fe- 
males to either abandon their nests or temporarily 
leave their nests to forage, allowing the eggs to 
chill and die. 

Adult Mortality. Mortality data for goshawks in the 
WGLR are based almost solely on females found 
killed at or near their nests (Erdman et al. 1998). 
Thus, there are no data available prior to our study 
on causes of goshawk mortality away from their nests 
or during the non-breeding season. Our estimate of 
annual survival (74 ± 7.8%) based on telemetry was 
quite similar to mark-recapture estimates in Califor- 
nia (61-69%; DeStefano et al. 1994), New Mexico 
(60-96%; Kennedy 1997), and northern Arizona 
(69-87%; Reynolds and Joy 1998). All of these au- 
thors indicate imprecision in their studies due to a 
variety of reasons, and DeStefano et al. (1994) con- 
cluded accurate estimates of survival based on mark- 
resightings would require large numbers of marked 
birds, high resighting rates, and a minimum of 5 yr 
of data. This robust a data set has not been and is 
unlikely to be collected in the WGLR. In contrast, 
White and Garrott (1990) indicated survival estimates 
based on radiotelemetry requires smaller samples in 
general than mark-resighting estimates. Further- 
more, backpack radio attachments appear to have no 
significant effect on survival of goshawks (Reynolds 


et al. 2004) . Our data supported White and Garrott’s 
(1990) assertion; we were able to estimate survival 
rates with reasonable precision through marking con- 
siderably fewer birds than banding and resighting 
would require. However, we did not have a sufficient 
sample of radio-tagged birds to estimate temporal 
and gender differences in adult survival. 

One male that died during the winter of 1999- 
2000 had been banded as a juvenile at Hawk Ridge, 
MN, during the fall migration of 1988 (D. Evans 
pers. comm.). This male and his mate had fledged 
two young successfully in 1999. To our knowledge 
this 11-yr male is the oldest known recorded breed- 
ing male goshawk reported for North America. In- 
terestingly, the oldest reported female goshawk (12 
yr old) in North America was also reported from 
Minnesota (Evans 1981). 

The majority of information on causes of mor- 
tality among adult goshawks is anecdotal (Squires 
and Reynolds 1997). Goshawks succumb to several 
different diseases and parasites (Redig et al. 1980, 
Ward and Kennedy 1996, C. Boal unpubl. data). 
Accidents and injuries, such as flying into windows 
(C. Boal unpubl. data) or choking on prey (Blox- 
ton et al. 2002), also result in mortality. The pri- 
mary documented cause of mortality among free- 
ranging goshawks, however, appears to be 
predation (Squires and Kennedy in press). 

Known predators of adult goshawks include 
Great Horned Owls (Rohner and Doyle 1992, Boal 
and Mannan 1994, Erdman et al. 1998), Golden 
Eagles {Aquila chrysaetos; Squires and Ruggiero 
1996), Pine Martens {Martes americana; Paragi and 
Wholecheese 1994), and fishers (Erdman et al. 
1998). Of five adult goshawks taken by predators 
in our study, two were killed by Great Horned 
Owls. In the WGLR, fishers may also be an impor- 
tant predator of goshawks; predation by fishers was 
identified as the cause of mortality for four adult 
female goshawks in Wisconsin (Erdman et al. 
1998) and two of five goshawk deaths in our study. 
We believe the goshawk killed in our study by a 
Red-tailed Hawk may be an exceptional incident, 
but the two species have been observed engaged 
in physical agonistic encounters (Crannell and 
DeStefano 1992, C. Boal unpubl. data). In areas of 
sympatry (La Sorte et al. 2004), Red-tailed Hawk 
predation may be more common. 

Most mortality data for goshawks is for the nest- 
ing season. We found that mortality occurred with 
equal frequency in the breeding and winter sea- 
sons, suggesting that survival outside of the breed- 


September 2005 


Biology 


227 


ing season is an important aspect of goshawk pop- 
ulation dynamics. Our data also suggested that, 
despite legal protection, persecution was still a fac- 
tor affecting goshawk survival. 

Results from Wisconsin (Erdman et al. 1998) 
and our study suggested predators were a major 
cause of goshawk mortality in the WGLR. However, 
the influence of predators on goshawk population 
demography and whether current predation rates 
are similar to historic rates or have increased as a 
consequence of human activities (e.g., timber har- 
vest, reintroduction of fishers) in the WGLR, as 
suggested by Erdman et al. (1998), has yet to be 
assessed rigorously. The development and use of 
standardized field methods for evaluating causes of 
mortality of goshawks and publication of existing 
mortality data would be helpful in this regard. 
Without reliable survival data, rates of population 
growth or decline cannot be estimated accurately 
for the WGLR goshawk population. 

Acknowledgments 

We appreciate our dedicated field assistants that en- 
dured long hours of work: L. Belmonte, W. Estes, J. Sam- 
mler, R. Sandstrom, and A. Wester. A. Roberson and B. 
Smithers assisted with practically all aspects of this study, 
and M. Solensky assisted with trapping. D. Harrington 
and A. Weaver provided nest sites, and members of the 
Minnesota Falconers’ Association assisted with surveys. 
Personnel from the many cooperating agencies and or- 
ganizations provided an array of assistance and logistical 
support during the projects. They include R. Baker, J. 
Blanchard, J. Casson, M. Cole, J. Gallagher, L. Grover, M. 
Hamady, J. Hines, M. Houser, G. Kirk, B. Marty, R. Vora, 
and A. Williamson. Funding for this project was provided 
by the Chippewa National Forest, The National Council 
for Air and Stream Improvement, Minnesota Department 
of Natural Resources, Potlatch Corporation, Leech Lake 
Band of Ojibway, Superior National Forest, U.S. Fish and 
Wildlife Service, The Raptor Center at the University of 
Minnesota, The Minnesota Falconers’ Association, Voya- 
geurs National Park, the Minnesota Cooperative Fish and 
Wildlife Research Unit, and grants from the National 
Forest Foundation and the National Fish and Wildlife 
Foundation. 

Literature Cited 

Bart, J. and D.S. Robson. 1982. Estimating survivorship 
when the subjects are visited periodically. Ecology 63: 
1078-1090. 

Bloxton, T.D., A. Rogers, M.F. Ingraldi, S. Rosen- 
stock, J.M. Marzluef, and S.P. Finn. 2002. Possible 
choking mortalities of adult Northern Goshawks. J. 
Raptor Res. 36:141—143. 

Boat, C.W. 1994. A photographic and behavioral guide 
to aging nestling Northern Goshawks. Stud. Avian 
Biol. 16:32-40. 


, D.E. Andersen, and P.L. Kennedy. 2001. Home 

range and habitat use of Northern Goshawks {Acciptter 
gentilis) in Minnesota. Minnesota Cooperative Fish 
and Wildlife Research Unit, St. Paul, MN U.S.A. 

, , AND . 2003. Home range and res- 
idency status of Northern Goshawks breeding in Min- 
nesota. Condor 105:811— 816. 

AND R.W. Mannan. 1994. Northern Goshawk diets 

in ponderosa pine forests on the Kaibab Plateau. Stud. 
Avian Biol. 16:97—102. 

Bull, E.L. andJ.H. Hohmann. 1994. Breeding biology of 
Northern Goshawks in northeastern Oregon. Stud 
Avian Biol. 16:103-105. 

Crannell, D. and S. DeStefano. 1992. An aggressive in- 
teraction between a Northern Goshawk and a Red- 
tailed Hawk. J. Raptor Res. 26:269-270. 

Daw, S.K., S. DeStefano, and R.J. Steidl. 1998. Does sur- 
vey method bias the description of Northern Goshawk 
nest-site selection? J. Wildl. Manag. 62:1379-1384. 

DeStefano, S.K. Daw, S.M. Desimone, and E.C. Meslow. 
1994. Density and productivity of Northern Goshawks: 
implications for monitoring and management. Stud. 
Avian Biol. 1 6 : 88-9 1 . 

, S., B. WOODBRIDGE, AND PJ- Detrich. 1994. Sur- 
vival of Northern Goshawks in the southern Cascades 
of California. Stud. Avian Biol. 16:133—136. 

Doyle, F.I. 1995. Bald Eagle {Haliaeetus leucocephalus) and 
Northern Goshawk (Accipiter gentilis) nests apparently 
preyed upon by a wolverine (s) , (Gulo gulo), in the 
southwestern Yukon Territory. Canad. Field-Nat. 109: 
115-116. 

Einarsen, A.S. 1956. Determination of some predator 
species by field signs. Oregon State Monogr. Stud. Zool 
10 . 

Erdman, T.C., D.F. Brinker,J.P. Jacobs, J. Wilde, andT.O. 
Meyer. 1998. Productivity, population trend, and sta- 
tus of Northern Goshawks, Accipiter gentilis atricapillus, 
in northeastern Wisconsin. Canad. Field-Nat. 112:17- 
27. 

Evans, D.L. 1981. Banding recoveries from Hawk Ridge. 
Prairie Nat. 32:18. 

Johnson, D.H. 1999. The insignificance of statistical sig- 
nificance testing. J. Wildl. Manag. 63:763—772. 

Kaplan, E.L. and P. Meier. 1958. Nonparametric esti- 
mation from incomplete observations./. Am. Statistical 
Assoc. 53:457-481. 

Kennedy, P.L. 1997. The Northern Goshawk {Accipiter gen- 
tilis atricapillus) : is there evidence of a population de- 
cline? J. Raptor Res. 31:95-106. 

. 2003. Northern Goshawk {Accipiter gentilis atricap- 
illus) : a technical conservation assessment. USDA For- 
est Service, Rocky Mountain Region, Species Conser- 
vation Project. Published as a web document by USFS 
at http://www.fs.fed.us/ r2/scp/species_assessments/ 
northern-goshawk/ northern_goshawk.pdf. 

Kostrzewa, a. and R. Kostrzewa. 1990. The relationship 
of spring and summer weather with density and 


228 


Boal et al. 


VoL. 39, No. 3 


breeding performance of the buzzard Buteo buteo, gos- 
hawk Accipiter gentilis, and kestrel tinnunculus. Ibis 

132:550-559. 

Lapinski, N.W. 2000. Habitat use and productivity of the 
Northern Goshawk in the upper peninsula of Michi- 
gan. M.S. thesis, Northern Michigan Univ., Mar- 
quette, MI U.S.A. 

La Sorte, F.A., R.W. Mannan, R.T. Reynolds, and T.G. 
Grubb. 2004. Habitat associations of sympatric Red- 
tailed Hawks and Northern Goshawks on the Kaibab 
Plateau./. Wildl Manag. 68:307-317. 

McClaren, E., P.L. Kennedy, and S.R. Dewey. 2002. Do 
some Northern Goshawk nest areas fledge more 
young than others? Condor 104:343-352. 

Minnesota Department of Natural Resources. 2004. 
Ecological classification system. Minnesota Depart- 
ment of Natural Resources, St. Paul, MN U.S.A. 
http:/ /www.dnr.state.mn.us/ecs/index.html 

Paragi, T.G. AND G.M. Wholecheese. 1994. Marten, Mar- 
ies americana, predation on a Northern Goshawk Ac- 
cipiter gentilis. Canad. Field-Natur. 108:81-82. 

Pollock, K.H., S.R. Winterstein, C.M. Bunck, and P.D. 
Curtis. 1989. Survival analysis in telemetry studies: 
the staggered entry design./. Wildl. Manag. 53:7-15. 

Redig, P.T., M.R. Fuller, and D.L. Evans. 1980. Preva- 
lence of Aspergillus fumigatus in free-living goshawks 
{Accipiter gentilis atricapillus) . J. Wildl. Diseases 16:169— 
174. 

Reynolds, R.T. and S.M. Joy. 1998. Distribution, territory 
occupancy, dispersal and demography of Northern 
Goshawks on the Kaibab Plateau, Arizona: final re- 
port. Heritage Project No. 194045. Arizona Game and 
Fish Department, Phoenix, AZ U.S.A. 

, , and D.G. Leslie. 1994. Nest productivity, 

fidelity, and spacing of Northern Goshawks in north- 
ern Arizona. Stud. Avian Biol. 16:106-113. 


, G.C. White, S.M. Joy, and R.W. Mannan. 2004. 

Effects of radiotransmitters on Northern Goshawks: 
do tailmounts lower survival of breeding males ? J. 
Wildl. Manag. 68:25-32. 

Rohner, C. and F.I. Doyle. 1992. Food-stressed Great 
Horned Owl kills adult goshawk: exceptional obser- 
vation or community process? /. Raptor Res. 26:261- 
263. 

Squires, J.R. and P.L. Kennedy. In press. Northern Gos- 
hawk ecology: an assessment of current knowledge 
and information needs for conservation and manage- 
ment. Studies Avian Biology. 

AND R.T. Reynolds. 1997. Northern Goshawks 

{Accipiter gentilis) . In A. Poole and F. Gill [Eds.] , The 
birds of North America, No. 298. The Academy of 
Natural Sciences. Philadelphia, PA, and The Ameri- 
can Ornithologists’ Union, Washington, DC U.S.A. 

and L.F. Ruggiero. 1996. Nest-site preference of 

Northern Goshawk in south-central Wyoming. /. 
Wildl. Manag. 60:170-177. 

United States Fish and Wildlife Service. 1998. Notice 
of 12-month finding on a petition to list the Northern 
Goshawk in the contiguous United States west of the 
100* meridian. Federal Register 63:35183-35184. 

Ward, J.M. and P.L. Kennedy. 1996. Effects of supple- 
mental food on size and survival of juvenile Northern 
Goshawks. Auk 113:200-208. 

White, G.C. and R.A. Garrott. 1990. Analysis of Wildlife 
Radio-tracking Data. Academic Press, San Diego, CA 
U.S.A. 

Zachel, C.R. 1985. Food habits, hunting activity, and 
post-fledging behavior of Northern Goshawks {Accip- 
iter gentilis) in interior Alaska. M.S. thesis, Univ. Alas- 
ka, Fairbanks, AK U.S.A. 

Received 8 February 2004; accepted 27 September 2004 

Guest Editor: Stephen DeStefano 


/. Raptor Res. 39(3):229-236 
© 2005 The Raptor Research Foundation, Inc. 

RELATIONSHIPS BETWEEN WINTER AND SPRING WEATHER AND 
NORTHERN GOSHAWK (ACCIPITER GENTILIS) REPRODUCTION 

IN NORTHERN NEVADA 

Graham D. Fairhurst^ and Marc J. Bechard 

Raptor Research Center, Department of Biology, Boise State University, Boise, ID 83725-1365 U.S.A. 

Abstract. — Ecological factors, such as weather, play important roles in raptor population dynamics. We 
used logistic and Poisson regression analyses to investigate relationships between late winter, spring, and 
early summer temperatures and precipitation and Northern Goshawk {Accipiter gentilis) breeding, failure, 
and productivity in northern Nevada from 1992-2002. We also examined weather data for possible 
patterns that could explain reported trends in goshawk reproduction. Declines in occupancy of nesting 
territories by breeding goshawks were related to colder February and March temperatures and increased 
April precipitation. Warmer April temperatures and decreased precipitation in April-July favored re- 
productive success. Of all significant weather variables, only February and March temperatures had 
significant temporal trends. Although adverse weather is known to affect goshawk reproduction by 
decreasing nestling growth and survival, it is unlikely that direct weather effects were responsible for 
reported reproductive trends in our study area. Weather may have operated indirectly, influencing 
reproduction through changes in goshawk hunting behavior or food supply. 

Key Words; Northern Goshawk, Accipiter gentilis; weather, breeding, population trends', Nevada. 


RELACIONES ENTRE EL CLIMA DE INVIERNO YDE PRIMAVERA Y LA REPRODUCCION DE AC- 
CIPITER GENTILIS EN EL NORTE DE NEVADA 

Resumen. — Factores ecologicos como el clima tienen un papel importante en la dinamica poblacional 
de las aves rapaces. Utilizamos analisis de regresion logistica y de Poisson para investigar las relaciones 
entre las temperaturas y las precipitaciones de fines del invierno, de la primavera y del comienzo del 
verano, y los fracasos o exitos reproductivos y la productividad de Accipiter gentilis en el norte de Nevada 
entre 1992 y 2002. Tambien examinamos los datos de clima para encontrar posibles tendencias que 
puedan explicar las tendencias documentadas de la reproduccion de estos halcones. La disminucion 
en la ocupacion de territorios de nidificacion por halcones reproductivos se relaciono con las tempe- 
raturas mas Mas de febrero y marzo y el aumento de las precipitaciones en abril. Las temperaturas mas 
calidas de abril y la disminucion de las precipitaciones en abril^ulio favorecieron el exito reproductivo. 
De todas las variables climaticas significativas, solo las temperaturas de febrero y marzo presentaron 
tendencias temporales significativas. A pesar de que es sabido que las condiciones climaticas adversas 
afectan la reproduccion de estos halcones al disminuir el crecimiento y la supervivencia de los polluelos, 
es poco probable que los efectos directos del clima fueran responsables de las tendencias reproductivas 
documentadas en nuestro sitio de estudio. Las condiciones climaticas pueden haber operado indirec- 
tamente, influenciando la reproduccion a traves de cambios en el comportamiento de caza de los 
halcones, o en la disponibilidad de alimento. 

[Traduccion del equipo editorial] 


Weather can directly influence Northern Gos- 
hawk {Accipiter gentilis) population dynamics by af- 
fecting survival (Zachel 1985, Squires and Reyn- 
olds 1997, Bloxton 2002), movements (Marcstrom 
and Kenward 1981, Squires and Ruggiero 1995), 
and nestling development (Kostrzewa and Kostrze- 


^ Corresponding author’s email address: gdfair@gmail. 
com 


wa 1990). Temperature and precipitation may also 
indirectly affect prey populations (Van Horne et al. 
1997, Bloxton 2002), foraging behavior (Zachel 
1985), and other mortality factors (Newton 1979). 

Studies addressing the relationships between 
weather and goshawk reproduction (Kostrzewa 
and Kostrzewa 1990, 1991, Patla 1997, Penteriani 
1997, Ingraldi 1998, Bloxton 2002), generally 
agree that colder and wetter spring weather nega- 


229 


230 


Fairhurst and Bechard 


VoL. 39, No. 3 


lively affects goshawk reproduction; however, the 
association between winter weather and goshawk 
reproduction is not well studied. These studies 
have also only considered time periods <6 yr (In- 
graldi 1998). 

We previously reported declines in goshawk nest- 
ing territory occupancy and increases in breeding 
failure in northern Nevada from 1992-2002 (Be- 
chard et al. in press). Determining the ecological 
factors responsible for these reproductive trends is 
difficult because any variable that showed a tem- 
poral trend from 1992-2002 will consequentially 
be correlated with reproduction. However, because 
of the known links between weather and goshawk 
reproduction and the abnormally low precipitation 
and drought conditions reported in the northern 
Great Basin from 1999-2002 (National Drought 
Mitigation Center 2003), we suspected that weath- 
er conditions affected goshawk reproduction in 
northern Nevada. Here, we address the associa- 
tions between late winter, spring, and early sum- 
mer temperature and precipitation and long-term 
trends in goshawk reproductive performance. 

Methods 

Study Area. We conducted the study in the Indepen- 
dence and Bull Run Mountain ranges of Elko County, 
northern Nevada, during 1992-2002. The study area ex- 
tended ca. 150 km north-to-south, 10-30 km east-to-west, 
and encompassed ca. 94 000 ha. The area is a mosaic of 
public lands administered by the United States Depart- 
ment of Agriculture Forest Service (Humboldt-Toiyabe 
National Forest) interspersed with private lands. Eleva- 
tions range from ca. 1700-3000 m on the highest peaks. 
A mixture of land uses occurred in the study area, in- 
cluding cattle ranching, gold mining, and outdoor rec- 
reation (hunting, camping, and off-road vehicle use). 

The sagebrush steppe was typified by big sagebrush 
{Artemisia tridentata) , bitterbrush {Purshia tridentata), and 
rabbitbrush {Chrysothamnus . Common native grass- 
es included native bluebunch wheatgrass {Pseudoroegneria 
spicata) and Idaho fescue {Festuca idahoensis), and intro- 
duced cheatgrass (Bromus tectorum) and medusahead 
wildrye (Taeniatherum caput-medusae) also occur (Loope 
1969; P. Jelinek, Humboldt-Toiyabe National Forest, pers. 
comm.). In the study area, goshawks nested exclusively 
in quaking aspen (Populus tremuloides) , which occurred in 
naturally-fragmented stands where sufficient moisture 
was present. Subalpine fir {Abies lasiocarpa) replaces quak- 
ing aspen above elevations of 2500 m (Loope 1969). 

Dense willow {Salix spp.), thickets, and cottonwoods 
{Populus spp.) occurred in riparian areas at all elevations. 

Field Methods. In April of 1991, 1992, and 1994-96, 
we used helicopters to initially locate and subsequently 
survey all historical goshawk nesting territories in the 
study area. We defined a nesting territory as the area 
containing one or more nests occupied by a single pair 
of goshawks in any breeding season (Postupalsky 1974, 


Woodbridge and Detrich 1994, Reynolds and Joy 1998) 
We discontinued helicopter surveys after 1996. 

Beginning in mid-May of each year from 1992-2002, 
we conducted ground surveys of all historically-occupied 
nesting territories (i.e., previously located by helicopter 
and used by goshawks) to determine occupancy by gos- 
hawk breeding pairs. We conducted ground surveys on 
foot and with all-terrain vehicles by returning to nesting 
territories and thoroughly searching stands and adjacent 
stands for the presence of breeding goshawks. Because 
territories in our study area were relatively small and nest 
structures obvious, we were able to search each territory 
completely and, therefore, assume a uniform probability 
of detection. We were unable to reach all nesting terri- 
tories each year because roads throughout the study area 
were periodically snow-covered or washed-out, and pri- 
vate land was not always accessible. 

Beginning in mid-June of each year, we revisited oc- 
cupied nesting territories to determine productivity (Be- 
chard et al., in press). We climbed all occupied nest trees 
and counted and banded nestlings when they were ap- 
proximately 21-31 d old (age based on nestling plumage, 
Boal 1994). We considered a pair failed if there was no 
sign of goshawks (adults or nestlings) at or near an oc- 
cupied nest when it was revisited in June. 

Data Analyses. We determined nesting territory occu- 
pancy, productivity per breeding pair, breeding failure, 
and productivity per successful pair. Because we visited 
most nests only twice during the breeding season, there 
was as much as a 30-d interval between our nest visits. 
Therefore, we could not always determine the cause of a 
nest failure. There were no instances that clearly indicat- 
ed the nesting attempt had failed due to depredation or 
any other factor unrelated to weather; therefore, we in- 
cluded all failures in our analysis. 

We downloaded weather data for the study area from 
the Natural Resources Conservation Service/SNOTEL 
website (Jack’s Creek Upper weather station; Natural Re- 
sources Conservation Service 2003). Because SNOTEL 
considered temperature data for 2002 unreliable, we did 
not include those data. 

To analyze 11-yr trends in reproduction, we used logis- 
tic regression for binary outcomes (i.e., nesting territory 
occupancy and nesting failures) and Poisson regression 
for counted outcomes (i.e., productivity). To account for 
the fact that repeated measurements on a nesting terri- 
tory over time might not be statistically independent (Al- 
lison 1999), we used Generalized Estimating Equations 
(“GEE;” PROC GENMOD; SAS Institute, Inc., Cary, NC 
U.S.A.), which allowed us to cluster our data by nesting 
territory (Stokes et al. 2000). 

We modeled mean daily temperature (°C), cumulative 
monthly precipitation (cm; Fig. 1,2), and year as contin- 
uous explanatory variables. We used only weather data 
from January-July, as we felt that they were most biolog- 
ically relevant. We considered individual months and 
groups of months (i.e., January and February, February 
and March) in analyses, resulting in 27 possible explan- 
atory variables per reproductive outcome (13 for mean 
daily temperature, 13 for cumulative monthly precipita- 
tion, and 1 for year). Due to low sample sizes {N = 11 
yr) , we did not consider more than one explanatory var- 
iable per model. Therefore, to avoid over specifying our 


September 2005 


Biology 


231 


0 

01 

£ 

3 

£ 

0) 

Q. 

E 

0) 


to 

Q 

c 

cd 

<]} 



Year 

Figure 1. Mean daily temperature for January-July in the Independence and Bull Run mountains, Nevada for 1992- 
2002. 


models, we ran separate univariate models for each ex- 
planatory weather variable and employed a pre-screening 
procedure to decide which models best explained the 
relationship between weather and goshawk reproduction. 

During pre-screening, we only considered weather var- 
iables that appeared to be biologically relevant. For ex- 
ample, July weather would not be biologically related to 
occupancy in the same year because occupancy occurs 
before July. In addition, for each of the four reproductive 
outcomes, we ran several competing univariate models: 
each competing model used weather data from a single 
month or group of months as the explanatory variable. 
For example, II competing occupancy models used 
mean daily temperature as the explanatory variable, and 
an additional II occupancy models used cumulative 
monthly precipitation. For each reproductive outcome, 
we then selected the models with the single most statis- 
tically significant temperature and precipitation variables. 
Thus, for each reproductive outcome, we presented 
three separate univariate models: one for temperature, 
one for precipitation, and one for year. To avoid inflated 
Type I error rates, we assessed significance of all models 
using a step-down Bonferroni correction (Holm 1979). 
Although our statistical method has the potential to pro- 
duce spurious results (Freedman 1983), the GEE has no 
other measure by which to assess multiple competing 
models. 

We evaluated the results of logistic regression models 
by exponentiation of the model coefficient to obtain 
odds ratios, and we evaluated the results of Poisson re- 
gression by exponentiation of the model coefficient to 
obtain percent increase in the mean values of dependent 
variables. In the analyses, we only included nesting ter- 
ritories for which we knew the reproductive outcome. 

To determine if any patterns existed in local weather 
variables, we used simple linear regression (JMP IN; SAS 
Institute, Inc., Cary, NC U.S.A.). For each weather vari- 
able that was significantly related to occupancy and fail- 


ure (one temperature variable and one precipitation var- 
iable per each reproductive outcome) , we regressed year 
against the temperature or precipitation variable. 

Results 

We initially located 27 nesting territories in 1992, 
and found five, five, and four additional nesting 
territories in 1993, 1994, and 1996, respectively. We 
monitored a mean of 32.5 ± 4.7 (x ± SD) nesting 
territories annually (Table 1). Goshawks occupied 
an average of 20.3 ± 6.7 of these nesting territories 
each year. The odds of nesting territory occupancy 
by breeding pairs increased by 55.8% with each 
1°C increase in combined February and March 
mean daily temperature (odds ratio = 1.558, P = 
0.018; Table 2). The odds of occupancy of a nest- 
ing territory increased by 7.7% for every cm in- 
crease in cumulative April precipitation (odds ratio 
= 1.077, P — 0.03). We detected a significant cool- 
ing trend in combined February and March mean 
daily temperature in our study area (r^ = 0.41, P 
— 0.04) , but we found no trend in cumulative April 
precipitation (i^ < 0.0002, P = 0.97). 

Goshawk breeding pairs fledged a mean of 2.27 
± 0.76 young annually. Mean productivity per 
breeding pair increased 10.5% for every 1°C in- 
crease in the mean daily April temperature (Table 
3; percent change in mean = 1.105, P < 0.0001). 
Mean productivity per breeding pair decreased by 
1.9% for every 1 cm increase in combined April 
and May precipitation (percent change in mean = 
0.981, P< 0.0003). 


232 


Fairhurst and Bechard 


VoL. 39, No. 3 


E 

o. 

c 

o 

'-4-» 

CO 

■q. 

o 

CD 

k. 

Q. 

x: 

c 

o 


.1 

3 

E 

3 

o 



Jan Feb Mar Apr May Jun Jul 


• 

1992 

o 

1993 

Y 

1994 

V 

1995 

■ 

1996 

□ 

1997 

♦ 

1998 

O 

1999 

A 

2000 

A 

2001 

• 

2002 


Month 


Figure 2. Monthly precipitation in the Independence and Bull Run Mountains, Nevada, for 1992—2002, expressed 
as cumulative monthly totals for January-July. 


On average, 13.5% of breeding attempts failed. 
The odds of failure decreased by 40.5% with every 
1°C increase in mean daily April temperature (Ta- 
ble 2; odds ratio = 0.595, P < 0.0007) and in- 
creased by 8.7% with each cumulative 1 cm in- 
crease in combined May and June cumulative 
precipitation (odds ratio = 1.087, P< 0.006). We 
found no trends in April (r^ = 0.094, P — 0.39) or 
combined May and June (r^ = 0.007, P = 0.80) 
temperatures. 

Successful pairs produced a mean of 2.64 ± 0.57 
young. For each 1°C increase in mean daily April 
temperature, we found a 6.4% increase in mean 
productivity per successful pair (Table 3; percent 
change in mean = 1.064, P < 0.0001). Mean pro- 
ductivity per successful pair decreased by 3.2% for 
each cumulative cm of combined June and July 
precipitation (percent change in mean = 0.978, P 
= 0.042). 

Discussion 

Long-term trends in goshawk reproduction were 
significantly related to weather, with a stronger in- 
fluence of temperature than of precipitation. Al- 
though late winter temperatures decreased in the 
study area from 1992—2002, our results suggested 
that warmer late winter temperatures favored gos- 
hawk breeding. Decreased productivity has been 
related to colder and wetter spring weather (Kos- 
trzewa and Kostrzewa 1990, Patla 1997, Penteriani 
1997, Bloxton 2002). Colder temperatures increase 


energetic stresses and increase dietary demands on 
raptors and can result in non-laying (Newton 
1979). Studies correlating winter temperatures 
with North American goshawk reproduction are 
lacking, but in European goshawks winter temper- 
atures were not related to occupancy by breeding 
pairs (Kostrzewa and Kostrzewa 1991). However, 
the larger size of European goshawks may make 
them more robust to temperature and energetic 
demands early in the breeding season than the 
smaller North American subspecies (Kendleigh 
1970). 

Our finding of increased April precipitation fa- 
voring occupancy by breeding pairs was unusual, 
and we found no previous studies to support this 
result. Moreover, increased precipitation early in 
the breeding season is typically associated with re- 
duced numbers of breeding pairs (Kostrzewa and 
Kostrzewa 1990, Ingraldi 1998, Bloxton 2002). Per- 
haps our finding of statistical significance does not 
necessarily relate to biological relevance, and the 
significant result is spurious. 

The temporal trend in the failure of breeding 
attempts was strongly related to April temperature 
and cumulative May and June precipitation. In- 
creased precipitation and decreased temperatures 
during the egg-laying and early nestling periods 
can increase egg and nestling mortality rates (Hog- 
lund 1964, Zachel 1985) and affect nestling devel- 
opment (Kostrzewa and Kostrzewa 1990). 


Table 1. Annual reproductive performance of Northern Goshawk nesting territories based on surveys conducted May-June 1992-2002 in the Indepi 


Biology 


233 


September 2005 


c 

u 


cn 

CO 

OJ 

t-i 


03 

u 

<u 

P3 


ii 

o 

<u 

;-i 

CO 

3 

O 

Si 

CX 

B 


ti 

Tj 

> 

lU 

z 

co" 

!=1 

a 

o 


a 


CP 

t 3 

§ 

(U 

u 

p 

<u 

T3 



p 3 

X 


O 

X 

o 

CM 

00 


CM 

(SO 

cc 

I> 

X 


1> 


Sd 

P 


1> 

00 

(J3 

03 

<o 


q 

q 


CM 

in 


q 

, 

Dh 

5 

CO 


d 

d 

d 

d 

d 

d 

d 

rH 

d 

rH 

d 


d 

0 

2 

0 

Z 

P 

CO 

M 

U 

U 

S 

+1 

o 

+1 

00 

+1 

CO 

+1 

CO 

+ 1 

rH 

+1 

+1 

fr 

+ 1 
J> 

+1 

03 

+ 1 
o 

+ 1 
o 

1 

+1 



P 


Oi 

oq 

I> 

in 

q 


CO 

rH 

q 

q 

m 


q 


|X 

tzs 

pel 

g 4 

CM 

CM 

CM 

CM 

CM 

ci 

ci 

ob 

(ci 

(ci 


ci 



ps; 

w 


CM 


CM 

O 


in 


CO 

X 

m 

o 


X 




C7i 


CM 


03 

00 

rH 

q 


q 



i> 


No. 

Pi 

o 

z 

p 

0 

z 

HH 

Q 

d 

+1 

+1 

00 

I— s 
+1 

nH 

+1 

d 

+1 

on 

d 

+1 

m 

rH 

+1 

CM 

rH 
+ 1 
CO 

d 
+ 1 
<03 

rH 

+ 1 
CM 

rH 

+ 1 
GO 

1 

d 

+1 



.0 

w 

r^ 

o 


oq 


o 

CM 

q 

q 

q 

q 


q 




3 

P3 

CN 

CM 

CM 

rH 

ci 

CM 

CM 

rH 

cb 

rH 

rH 


(ci 

















X 



d 

2 

•o 

0 













00 

rH 



X 

Z 

t-h 

o 

J> 

X 

<n 

03 

o 

CO 

rH 

m 

o 

X 

+ 1 




P 

X 

in 


-cf 


CO 

'vh 

CM 

CO 

CM 

rH 

1> 



P 

0 













'Cf^ 

q 



H 














GO 



S 

pi 

u 

X 

in 

in 

q 

G 

q 



O 

rH 

q 


q 



X 



CM 

d 

00 

<o 

in 

cb 

<03 

d 

GO 

ci 

1 

GO 






t-H 


CM 



rH 

CM 


CM 

•rri 


rH 



(X 

















X 

















p 














X 



bh 

CO 














cb 


d 

c« 

03 

rH 

CM 


o 

00 

03 

in 

CM 

X 

o 


GO 

+1 


2 

X 

o 


CM 

CM 

CM 

CM 

CM 


rH 

rH 

rH 

rH 


(03 



U 

IX 












rH 

q 



P 

















C/3 














rH 



0 














rH 



z 

rC 

C/3 













cb 


0 

Q 

p 3 

CM 


(03 

in 

o 

(03 

00 

i> 

X 

CO 

i> 

CM 

+1 


2 



CM 

CM 

X 

CM 

00 

rH 

i—S 

rH 

rH 

rH 


rH 


§ 













CM 

q 



PQ 














<03 

















rH 


































X 



U 














x' 



< 


in 

S-H 

on 

q 

CM 

q 

X 

q 

<o 

q 

CO 

1 

^H 



Ph 

P 

u 


1— ( 

00 

d 

(m 

(m 

(SO 

'Cf 

f — ( 


ob 

CM 

1 

+1 




00 

J> 




00 

m 

m 

in 

-'f 

CM 


'Cf 



u 

0 














ci 

CO 


o 

CO 
















z 

W 

Q 
















S 

X 

n 













cb 


C/3 

w 

0 

X 

p 

u 

u 

CM 

in 

X 

i> 

o 

o 

00 

j> 

X 

X 


X 

+1 

GO 


2 


CM 

CM 

CM 

CM 

eo 

CM 

rH 

I — s 

rH 

rH 


CM 

CM 


d 

X 

0 













d 


2 

H 














CM 



c/3 
















0 

X 

s 

0 

s 













1> 


z 















0 

P 

§ 

x> 

CM 

1> 

i> 

t-H 


CO 

CO 

X 

o 

rH 

X 

+ 1 
m 
ci 

2 

u 

2 

X 

p< 

p 

c/5 

CM 

on 

on 

en 

•'f 

CM 

(30 

CO 

eo 

oo 

GO 

m 

X 



H 














GO 





CM 

SSO 


in 

CO 

1> 

X 

03 

O 

rH 

CM 

23 

u 

c 

rt 





CJ3 

03 

(03 

<03 

(03 

(03 

03 

03 

o 

o 

O 

r® 




03 

03 

<03 

03 

(03 

03 

03 

(03 

o 

o 

o 






lH 



rH 

1^ 

rH 

i-H 


CM 

CM 

CM 



T3 

u 

a 

E 

Si 

a 

11 

T3 

w 

O 

S5 


bsD 

. s 

T3 H 
U 03 


I 

£r- 

;s 

<J 

zs 

T5 

S 

a. 

II 
Si 
SU 

XI 


2 

U a, 
D "O 


I 

u 

3 

X 

0 

Si 

S3- 

X 

u 

1 

a 


Al 


o3 

o- 

<U 

'S 

- s 

-S (J 

+j 

c2 '? 

^ I 

8 a 

3 ^ 


X 

a 

3 


SJ 

V 

X 

3 


bo 


c 

u 

'c3 


a 

Cl. 

bo 

eft 

.s 



T3 

V 

qj 


G 

u 

CC 

Xi 

QJ 

^H 

TJ 

o 

eft 

4-* 

eft 

G 

Qj 

V 

U 

CJ 

U 

u 

C 

V 


CL, 

£ 

u 


234 


Fairhurst and Bechard 


VoL. 39, No. 3 


Table 2. Odds ratios, confidence intervals (Cl), and significance of weather variables related to Northern Goshawk 
nesting territory occupancy in the Independence and Bull Run Mountains, Nevada, for 1992—2002. 


Model Term 

Odds Ratio 

95% CP 

P-value^ 

Occupancy by breeding pairs 
Year‘s 

0.785 

0.701-0.880 

<0.0002 

Mean daily combined February and March 
temperature 

1.558 

1.135-2.138 

0.018 

Cumulative April precipitation 

1.077 

1.014-1.143 

0.03 

Failure 

Year 

1.157 

1.021-1.312 

0.044 

Mean daily April temperature 

0.595 

0.455-0.780 

<0.0007 

Cumulative combined May and June 
precipitation 

1.087 

1.031-1.146 

<0.006 


^ Odds ratios are nonsignificant if confidence interval covers 1.0 (even odds). 
Significance of terms assessed using a step-down Bonferroni adjustment (Holm 1979). 
Data for year model terms taken from Bechard et al. (in press). 


Despite evidence that weather can directly affect 
goshawk breeding, it is unlikely that direct weather 
effects are solely responsible for our reported 
trends in reproduction (Newton 1998). We found 
no significant temporal trends in weather related 
to nest failure, suggesting changes in that repro- 
ductive variable were due to other factors such as 
reduced hunting and food provisioning due to 
continued rainfall (Zachel 1985, Bloxton 2002). 
Also, depredation of goshawk nests can result in 
nest failures. Because our nest visits were several 
weeks apart, and we could not determine the exact 
cause of nest failure in all cases, we included all 


failures in our analysis, possibly biasing our results. 
Nevertheless, we found no direct evidence indicat- 
ing that other factors unrelated to weather, such as 
depredation by Great Horned Owls {Bubo vir^ni- 
anus), played a significant role in the breeding fail- 
ures we observed. 

The confounding influences of unmeasured, but 
plausible, factors that may have changed during 
the study period complicated the analysis. Obvious 
among these was a possible trend in prey popula- 
tions. Goshawks respond numerically to changes in 
numbers of prey (McGowan 1975, Doyle and Smith 
1994). In our study area, they relied heavily on 


Table 3. Percent change in mean, confidence intervals (Cl), and significance of weather variables related to North- 
ern Goshawk productivity in the Independence and Bull Run Mountains, Nevada, for 1992-2002. 


Model Term 

Percent Change 
IN Mean 

95% CP 

P-value’^ 

Productivity per breeding pair 
Year‘s 

0.985 

0.963-1.008 

0.20 

Mean daily April temperature 

1.105 

1.070-1.140 

<0.0001 

Cumulative combined April and May 
precipitation 

0.981 

0.971-0.990 

<0.0003 

Productivity per successful pair 
Year 

1.003 

0.989-1.017 

>0.90 

Mean daily April temperature 

1.064 

1.040-1.089 

<0.0001 

Cumulative combined June and July 
precipitation 

0.978 

0.962-0.994 

0.042 


“‘Percent changes in the means are nonsignificant if confidence interval covers 1.0 (even odds). 
•’ Significance of terms assessed using a step-down Bonferroni adjustment (Holm 1979). 

Data for year model terms taken from Bechard et al. (in press). 


September 2005 


Biology 


235 


Belding’s ground squirrels {Spermophilus beldingi) 
for food (Younk and Bechard 1994, Younk 1996), 
but because we did not census ground squirrels, 
we could not determine what effect change in 
ground squirrel populations had on goshawk re- 
production. 

Further, interactions of weather and prey abun- 
dance may affect raptor reproduction (Gargett et 
al. 1995, Steenhof et al. 1997, Bloxton 2002). 
Weather has been shown to affect ground squirrel 
populations in other parts of the northern Great 
Basin (Van Horne et al. 1997). Bloxton (2002) at- 
tributed increased breeding failure and signifi- 
cantly lower productivity to reduced abundances of 
goshawk prey species following the wet and cold 
winter and spring of a La Nina weather event in 
western Washington. He noted that goshawks did 
not breed if weather had affected prey popula- 
tions. Although the climate of the northern Great 
Basin differs markedly from western Washington, 
the interactive effects of weather and prey may not. 

Acknowledgments 

This project was supported by the Nevada Department 
of Wildlife, U.S. Department of Agriculture Forest Ser- 
vice (Humboldt-Toiyabe National Forest), Independence 
Mining Company (now AngloGold-Meridian Jerritt Can- 
yon Joint Venture), and Boise State University. We would 
like to thank P. Bradley of the Nevada Department of 
Wildlife for his foresight in initiating this much-needed 
study of goshawks in an unusual habitat. We are deeply 
grateful to Forest Service biologists W. Amy, S. Anderson, 
P. Jelinek, J. Warder, and B. Whalen for all their efforts 
through the years to help with this project. Without the 
support of Independence Mining Company this project 
would never have been possible. We would like to espe- 
cially thank J. Bokich, J. Campbell, L. Gionet-Sheffield, 
G. Goodrich, M. Jones, S. Lewis, K. McAdoo, andj. Parks 
at the mine for their support. G. Fairhurst was supported 
by a Dan Montgomery Graduate Research Fellowship 
while at Boise State University. We are indebted to J. 
Younk, M. Shipman, and H. Smith for their substantial 
contributions to this project, and to L. Bond for her sta- 
tistical help. We would like to thank all the field assistants 
who contributed to this project. We are grateful to re- 
viewers P. Beier, C. Boal, and M. Goldstein for providing 
valuable comments and suggestions. Finally, we would 
like to thank the organizers of the goshawk symposium, 
Editors C. Boal and J. Bednarz for their hard work. 

Literature Cited 

Allison, P.D. 1999. Logistic regression using the SAS® 
system; theory and application. SAS Institute, Inc., 
Cary, NC U.S.A. 

Bechard, M.J., G.D. Fairhurst, and G.S. Kaltenecker. 
In press. Occupancy, productivity, turnover, and dis- 


persal of Northern Goshawks in portions of the north- 
eastern Great Basin. Stud. Avian Biol.\\n press. 

Bloxton, T.D., Jr. 2002- Prey abundance, space use, de- 
mography, and foraging habitat of Northern Gos- 
hawks in western Washington. M.S. thesis, Univ. Wash- 
ington, Seattle, WA U.S.A. 

Boat, C.W. 1994. A photographic and behavioral guide 
to aging nestling Northern Goshawks {Acdpitenr gentil- 
is). Stud. Avian Biol. 16:32-40. 

Doyle, F.I. and J.M.N. Smith. 1994. Population responses 
of Northern Goshawks to the 10-year cycle in num- 
bers of snowshoe hares. Stud. Avian Biol. 16:122—129. 

Freedman, D.A. 1983. A note on screening regression 
equations. Am. Statistician 

Gargett, V., E. Gargett, and D. Damania. 1995. The 
influence of rainfall on Black Eagle breeding over 31 
years in the Matopo Hills, Zimbabwe. 66:114— 

12L 

Hoglund, N.H. 1964. The hawk Accipiter gentilislAnn^ in 
Fennos Kandia (English translation). Viltrevy 2:195- 
269. 

Holm, S. 1979. A simple sequentially rejective multiple 
test procedure. Scand. J. Statist. 6:65-70. 

INGRAI.DI, M.F. 1998. Population biology of Northern 
Goshawks in east-central Arizona. Tech. Rep. 133, 
Nongame and Endangered Wildlife Program, Arizona 
Game and Fish Department, Phoenix, AZ U.S.A. 

Kendleigh, S.C. 1970. Energy requirements for existence 
in relation to size of bird. Condor 72:60—65. 

Kostrzewa, a. and R. Kostrzewa. 1990. The relationship 
of spring and summer weather with density and 
breeding performance of the buzzard Buteo buteo, gos- 
hawk Accipiter gentilis and kestrel Falco tinnunculus. Ibis 
132:550-559. 

AND . 1991. Winter weather, spring and 

summer density, and subsequent breeding success of 
Eurasian Kestrels, Common Buzzards, and Northern 
Goshawks. Auk 108:342-347. 

Loope, L.L. 1969. Subalpine and alpine vegetation of 
northeastern Nevada. Ph.D. dissertation, Duke Univ., 
Durham, NC U.S.A. 

Marcstrom, V. AND R. Kenward. 1981. Movements of 
wintering goshawks in Sweden. Swed. Wildl. Res. 12:1- 
36. 

McGowan, J.D. 1975. Distribution, density, and produc- 
tivity of goshawks in interior Alaska: final report. Alas- 
ka Department of Fish and Game, Juneau, AK U.S.A. 

National Drought Mitigation Center, [online]. 2003. 
Drought monitor archive (http://drought.unl.edu/ 
dm/archive. html) . 

Natural Resources Conservation Service, [online]. 
2003. Nevada/Utah Snotel Sites (http://www.wrcc. 
dri.edu/snotel/snogtb.html) . 

Newton, I. 1979. Population ecology of raptors. Buteo 
Books, Vermillion, SD U.S.A. 

. 1998. Population limitation in birds. Academic 

Press, New York, NY U.S.A. 


236 


Fairhurst and Bechard 


VoL. 39, No. 3 


Patla, S.M. 1997. Nesting ecology and habitat of the 
Northern Goshawk in undisturbed and timber harvest 
areas on the Targhee National Forest, greater Yellow- 
stone ecosystem. M.S. thesis, Idaho State Univ., Po- 
catello, ID U.S.A. 

Penteriani, V. 1997. Long-term study of a goshawk 
breeding population on a Mediterranean mountain 
(Abruzzi Apennines, central Italy) : density, breeding 
performance and diet./. Raptor Res. 31:308-312. 

PosTUPALSKY, S. 1974. Raptor reproductive success: some 
problems with methods, criteria, and terminology. In 
F.N. Hamerstrom, Jr., B.E. Harrell, and R.R. Olen- 
dorff [Eds.], Management of raptors. Raptor Re- 
search Foundation, Vermillion, SD U.S.A. 

Reynolds, R.T. and S.M. Joy. 1998. Distribution, territory 
occupancy, dispersal, and demography of Northern 
Goshawks on the Kaibab Plateau, Arizona. Final Rep. 
Arizona Game and Fish. Heritage Project No. 194045. 
Phoenix, AZ U.S.A. 

Squires, J.R. and R.T. Reynolds. 1997. Northern Gos- 
hawk (Accipiter gentilis ) . In A. Poole and F. Gill [Eds.] , 
The birds of North America, No. 298. The Academy 
of Natural Sciences, Philadelphia, PA and The Amer- 
ican Ornithologists’ Union, Washington, DC U.S.A. 

and L.F. Ruggiero. 1995. Winter movements of 

adult Northern Goshawks that nested in southcentral 
Wyoming./. Raptor Res. 29:5-9. 

Steenhof, K., M.N. Kochert, and T.L. McDonald. 1997. 


Interactive effects of prey and weather on Golden Ea- 
gle reproduction./ Anim. Ecol. 66:350—362. 

Stokes, M.E., C.S. Davis, and G.G. Koch. 2000. Categor- 
ical data analysis using the SAS system. SAS Institute, 
Cary, NC U.S.A. 

Van Horne, B., G.S. Olson, R.L. Schooley, J.G. Corn, 
AND K.P. Burnham. 1997. Effects of drought and pro- 
longed winter on Townsend’s ground squirrel demog- 
raphy in shrubsteppe habitats. Ecol. Monogr. 67:295- 
315. 

Woodbridge, B. and PJ. Detrich. 1994. Territory occu- 
pancy and habitat patch size of Northern Goshawks 
in the southern Cascades of California. Stud. Avian 
Biol. 16:83-87. 

Younk, J.V 1996. Breeding ecology of the Northern Gos- 
hawk in relation to surface gold mining in naturally- 
fragmented aspen forests of northern Nevada. M.S 
thesis, Boise State Univ., Boise, ID U.S.A. 

and M.J. Bechard. 1994. Breeding ecology of the 

Northern Goshawk in high-elevation aspen forests of 
northern Nevada. Stud. Avian Biol. 16:119-121. 

Zachel, C.R. 1985. Food habits, hunting activity, and 
post-fledging behavior of Northern Goshawks {Accip- 
iter gentilis) in interior Alaska. M.S. thesis, Univ. of 
Alaska Fairbanks, Fairbanks, AK U.S.A. 

Received 2 February 2004; accepted 9 March 2005 

Associate Editor: Michael I. Goldstein 


/. Raptor Res. 39(3);237-246 
© 2005 The Raptor Research Foundation, Inc. 


PATTERNS OF TEMPORAL VARIATION IN GOSPIAWK 
REPRODUCTION AND PREY RESOURCES 

Susan R. Salafsky^ and Richard T. Reynolds 

USDA Forest Service, Rocky Mountain Research Station, 2150 Centre Avenue, Building A, Suite 350, 

Fort Collins, CO 80526 U.S.A. 

Barry R. Noon 

Department of Fishery and Wildlife Biology, Colorado State University, Fort Collins, CO 80523 U.S.A. 

Abstract. — To investigate whether Northern Goshawk {Accipiter gentilis) reproduction is food-limited, 
we evaluated the reproductive output from 401 goshawk breeding opportunities on the Kaibab Plateau, 
Arizona during 1999-2002. Concurrently, we estimated densities of 10 goshawk prey species (seven birds, 
three mammals) using distance sampling. We then assessed the relationship between goshawk produc- 
tivity (number of fledglings produced) and prey density within and among years by relating the contri- 
bution of individual prey species and total prey density to goshawk productivity. We also estimated the 
proportion of total diet and biomass for each species that contributed &3% of all prey items. Total prey 
density was highly correlated with variation in goshawk productivity {F — 0.98, P = 0.012). Red squirrel 
( Tamiasciurus hudsonicus) density explained more variation in goshawk productivity than any other spe- 
cies (r^ = 0.94, P — 0.031), but density could not be estimated for every predominant prey species in 
goshawk diets. However, only red squirrels had a positive and significant relationship to goshawk pro- 
ductivity in terms of frequency {F = 0.97, P — 0.014) and biomass = 0.95, P = 0.033). Northern 
Flickers {Colaptes auratus) and cottontail rabbits (Sylvilagus spp.), which contributed the greatest fre- 
quency and biomass, respectively, to goshawk diets, showed no relationship with goshawk productivity. 
Even though goshawks on the Kaibab Plateau have a diverse diet and will readily switch to alternate 
prey species, goshawk productivity showed significant interannual variation. Our results suggest that the 
magnitude of goshawk productivity was determined by total prey density and annual variation was driven 
by differences in the densities of critical prey species. 

Key Words; Northern Goshawk; Accipiter gentilis; diet, distance sampling, predator-prey dynamics; prey density; 
productivity. 


PATRONES DE VARIACION TEMPORAL DE LAS PRESAS Y DE LA REPRODUCCION 
DE ACCIPITER GENTILIS 

Resumen. — Para investigar si la reproduccion de Accipiter gentilis se encuentra limitada por la dispo- 
nibilidad de alimento, evaluamos el rendimiento reproductive de 401 oportunidades reproductivas de 
estos halcones en Kaibab Plateau, Arizona, entre 1999 y 2002. A1 mismo tiempo, estimamos las densi- 
dades de 10 especies de presas para los halcones (siete aves, tres mamiferos) utilizando el metodo de 
conteo con distancias variables. Luego determinamos la relacion entre la productividad de los halcones 
(numero de volantones producidos) y la densidad de presas dentro y entre anos, relacionando la con- 
tribucion de cada especie de presa y la densidad total de presas con la productividad de los halcones. 
Tambien estimamos la proporcion de la dieta total para cada especie que contribuyo mas del 3% de 
todas las presas en la dieta. La densidad total de las presas se correlaciono fuertemente con la variacion 
en la productividad de los halcones {F = 0.98, P = 0.012). La densidad de la ardilla Tamiasciurus 
hudsonicus explico la mayor parte de la variacion en la productividad de los halcones con relacion a las 
otras especies {F = 0.94, P = 0.031), pero no se pudo estimar la densidad de cada especie de presa 
predominante en la dieta de los halcones. Sin embargo, solo la ardilla T. hudsonicus presento una 
relacion positiva y significativa con la productividad de los halcones en terminos de frecuencia {F = 
0.97, P = 0.014) y biomasa {F = 0.95, P — 0.033). Las aves del genero Colaptes y los conejos, que 
contribuyeron la mayor frecuencia y biomasa de la dieta de los halcones, respectivamente, no se cor- 


^ Corresponding author’s email address: salafsky@lamar.colostate.edu 


237 


238 


Salafsky et al. 


VoL. 39, No. 3 


relacionaron con la productividad de los halcones. A pesar de que los halcones en Kaibab Plateau tienen 
una dieta diversa y pueden cambiar facilmente a especies de presas alternativas, su productividad mostro 
una variacion interanual significativa. Nuestros resultados sugieren que la magnitud de la productividad 
de los halcones fue determinada por la densidad total de presas y que la variacion anual fue producida 
por las diferencias en la densidad de especies de presas criticas. 

[Traduccion del equipo editorial] 


To understand temporal variation in population 
size, it is necessary to focus on the factors that limit 
demographic processes, such as reproduction and 
survival. Ultimately, the availability of essential re- 
sources within a habitat regulates population 
growth. Resource availability, specifically food, is 
hypothesized to be an important limiting factor of 
many raptor populations (Newton 1979). Varia- 
tions in food supply often result in extensive fluc- 
tuations in population demographic parameters 
(Gotelli 1998, Newton 1998), but the mechanisms 
of food-limitation are difficult to quantify, espe- 
cially in complex systems. Consequently, most in- 
formation on the influence of food resources on 
population dynamics comes from correlations be- 
tween reproduction and food abundance (Martin 
1987). 

The magnitude of the effects of food-limitation 
on reproduction is poorly understood, especially 
for predators with broad diets, such as Northern 
Goshawks {Accipiter gentilis). Goshawks regularly 
consume a variety of prey including ground and 
tree squirrels, rabbits, medium to large passerines, 
woodpeckers, and gallinaceous birds (Squires and 
Reynolds 1997). The diversity of prey in their diets 
ultimately depends on the abundance and avail- 
ability of the local bird and mammal fauna, which 
varies geographically. In Canada, although gos- 
hawks regularly consumed several prey species 
(>5) , goshawk reproduction showed a strong func- 
tional response to only one species — snowshoe 
hare {Lepus amencanus; Doyle and Smith 2001). In 
contrast, 14 species of birds and mammals regular- 
ly contributed to goshawk diets in the southwestern 
United States (Reynolds et al. 1992). This diet di- 
versity may stabilize their breeding rates. When 
prey populations vary asynchronously, the ability of 
goshawks to switch between alternative prey species 
may result in less annual variation in reproduction 
than in areas where goshawks rely primarily on cy- 
clic populations of a single prey species (Newton 
1979). 

Our objectives were to: (1) determine if prey re- 
sources limit the reproductive rates of goshawks 
with relatively diverse diets and (2) describe how 


changes in prey populations may influence gos- 
hawk productivity (number of fledglings pro- 
duced) . If food is a limiting factor of goshawk pro- 
ductivity, then variation in the number of 
fledglings produced should be associated with fluc- 
tuations in prey resources. However, if there is a 
difference in the contribution of individual prey 
species, then goshawk productivity should respond 
to fluctuations in the densities of individual prey 
species. Finally, if the densities of important prey 
species vary in synchrony, then goshawk productiv- 
ity should exhibit greater temporal variation. To 
explore these relationships we studied goshawk 
productivity and prey resources on the Kaibab Pla- 
teau, Arizona during 1999-2002. 

Study Area 

The Kaibab Plateau is a large (95 X 55 km) forested 
island, surrounded by shrub-steppe desert, in northern 
Arizona. Steep slopes and escarpments form the eastern, 
southern, and western edges of the Kaibab Plateau and 
create a distinct boundary between the shrub-steppe des- 
ert at 1750 m elevation above sea level and the plateau 
(maximum elevation 2800 m). The northern edge of the 
plateau gradually descends to sagebrush desert, forming 
an indistinct boundary between the two landforms. 

The study area (1285 km^) on the Kaibab Plateau in- 
cluded forests above 2182 m elevation on the North Kai- 
bab Ranger District of the Kaibab National Forest. Four 
forest types dominated the study area: Pinyon-juniper {Fi- 
nns edulis-Juniperus spp.) woodlands occupied 106 km^ at 
lower elevations, ponderosa pine {Finns ponderosa) forests 
occupied 714 km^ at mid-elevation zones, mixed conifer 
{Abies concolor, Finns ponderosa, Fseudotsuga menziesii, Ficea 
engelmannii) forests occupied 275 km^ at the highest el- 
evations, and quaking aspen {Fopulns tremuloides) forests 
occupied 112 km^ interspersed among the other forest 
types (Joy 2002). 

Methods 

Goshawk Productivity. We estimated annual goshawk 
productivity per territory in 1999-2002. A territory was 
defined as the area (approximately 1 1 km^) defended by 
a pair of goshawks during the breeding season (Reynolds 
et al. 2005). Because goshawks may use more than one 
nest within a territory among breeding years (Reynolds 
et al. 2005), all nest structures were visited annually in 
spring to determine the territory occupancy status. If an 
active nest (nest containing eggs or young) was not lo- 
cated within an existing territory, we conducted system- 
atic surveys until we found an active nest or thoroughly 


September 2005 


Biology 


239 


searched the entire territory, which required a minimum 
effort of 10 person-days (Reynolds et al. 2004) . Each year, 
we also conducted surveys throughout the study area to 
locate territories not detected in previous years (Reyn- 
olds and Joy 2005). To determine nest status and fledg- 
ling production, all active nests were visited weekly 
throughout the breeding season. Goshawk offspring were 
counted in the nest 7-10 d prior to fledging or from the 
ground after fledging. Goshawk productivity was estimat- 
ed annually as the mean number of fledglings produced 
per territory under study. 

Prey Density. To obtain estimates of prey density, we 
conducted distance sampling (Buckland et al. 1993) 
along line transects from 1999-2002. Sixty 500-m tran- 
sects were placed randomly throughout the study area 
within two strata defined by the forest types (mixed co- 
nifer, ponderosa pine) that occupied most of the study 
area. We established 30 transects per stratum and char- 
acterized each transect by its elevation, tree species com- 
position, and tree density. Within a given year, we sam- 
pled each transect during three time periods that 
corresponded with specific goshawk reproductive stages: 
spring (28 May-24 June)-incubation/hatching stage, 
summer (25 June-22 July) -nestling stage, and late sum- 
mer (23 July-14 August) -fledging stage. To reduce travel 
time between transects and to increase sampling efficien- 
cy, transects were grouped by location. Transects were 
sampled in groups of four per day, and the sampling or- 
der of groups was determined using a random number 
table. Daily sampling began 0.5 hr after sunrise and was 
completed within 3 hr. All transects were sampled by one 
observer (Salafsky) during the 4 yr of the study. Sampling 
was not conducted during inclement weather (rain, 
winds >20 kph) due to reduced probability of prey de- 
tection. Prey seen or heard during sampling were iden- 
tified to species, and the perpendicular distance from the 
detected animal to the transect line was measured with a 
laser rangefinder (accurate to ±1 m). Data were collect- 
ed on 15 prey species common in goshawk diets on the 
Kaibab Plateau (S. Salafsky unpubl. data) and considered 
important components of goshawk diets in the south- 
western United States (Reynolds et al. 1992) . 

Goshawk Diet. The species composition of goshawk di- 
ets was determined from prey remains (pelage, plumage, 
skeletal parts) that were collected from active goshawk 
nest sites during weekly visits throughout the breeding 
season. Prey remains were pooled by territory and date 
collected, identified to species, and paired to assess the 
minimum number of individuals consumed (Reynolds 
and Meslow 1984) . The biomass contribution of individ- 
ual prey was based on the published mass of each avian 
(Dunning 1993) and mammal (Hoffmeister 1986) spe- 
cies. All methods for quantifying raptor diets have inher- 
ent biases (Marti 1987). However, Kennedy (1991) re- 
ported that estimates of prey use were similar for prey 
remain, pellet, and direct observation methods of diet 
analysis for goshawks in New Mexico. 

Data Analysis. We based goshawk productivity on the 
number of fledglings produced per territory under study. 
We classified territories based on ^1 attempt to breed on 
the territory, the identity of the adult birds, and the av- 
erage inter-territory distance (Reynolds et al. 2005). A 
high density of territories, a tendency of individuals to 


retain the same territory for life, and a delayed age at 
first breeding (Wiens and Reynolds 2005) suggests that 
the breeding habitat on the Kaibab Plateau was saturated. 
This evidence combined with the observed patterns of 
territory occupancy for individual adults over a 14-yr pe- 
riod (S. Salafsky unpubl. data) indicated that goshawks 
occupied the territories, even when we found little evi- 
dence of birds present. By including all territories rather 
than only those that were confirmed “active” or “occu- 
pied,” we accounted for all potential breeding opportu- 
nities and the full range of variability in the reproductive 
quality of territories. 

Variable distance sampling data were analyzed with 
program DISTANCE, Version 3.5 (Thomas et al. 1998) 
Reliable estimates of density from distance sampling de- 
pend on several critical assumptions: all individuals on 
the transect line were detected, all individuals were de- 
tected at their initial location, and all distances were mea- 
sured accurately (Buckland et al. 1993). Data collection 
methods were designed to meet these assumptions. Be- 
cause variable distance sampling uses a detection func- 
tion that compensates for differences in detection prob- 
abilities among species, habitats, and distances from 
transects (Emlen 1971, Buckland et al. 1993), density es- 
timates based on distance data are not confounded by 
factors affecting detectability and thus are representative 
of the true population size. Prey densities were estimated 
separately for mixed conifer and ponderosa pine to ac- 
count for differences in detection probabilities among 
forest types. These estimates were then multiplied by the 
proportion of each forest type within the study area and 
added together to calculate prey densities for the entire 
study area. Annual density estimates were computed only 
for species with sufficient sample sizes. Total prey density 
was calculated as the sum of the individual prey densities 
for species with a sufficient number of detections. We 
stratified total prey density by sampling period within 
each year to estimate prey densities associated with gos- 
hawk breeding phenology. 

We used the Tukey-Kramer adjustment for multiple 
comparisons of means to test for differences in goshawk 
productivity among years (PROG GLM, SAS Institute 
1999). Z-statistics were used to test for differences in 
mean prey densities among years and sampling periods 
(Buckland et al. 1993). To control for Type I error, we 
only tested for differences in density between specific 
pairwise comparisons (e.g., years of highest and lowest 
density) . To assess the relationship between goshawk pro- 
ductivity and prey density, we used linear regression 
(PROG REG, SAS Institute 1999), where annual goshawk 
productivity was the dependent variable, and estimates 
for individual prey species and summed over prey species 
were used as explanatory variables. Linear regression was 
also used to assess the relationship between goshawk pro- 
ductivity and prey species in the diet. In these regres- 
sions, annual goshawk productivity was the dependent 
variable and percent of total diet or biomass contribution 
for individual prey species were assessed as explanatory 
variables. We used an information-theoretic approach 
(Burnham and Anderson 2002) to identify the prey var- 
iables that explained the most annual variation in gos- 
hawk productivity per territory. A priori candidate models 
were developed to represent the potential effects of prey 


240 


Salafsky et al. 


VoL. 39, No. 3 


1.5 n 



1999 2000 2001 2002 

Year 


Figure 1. Mean number of Northern Goshawk fledglings produced per territory (±SE) on the Kaibab Plateau, 
Arizona, 1999-2002. 


density on goshawk productivity. We hypothesized that 
goshawk productivity would be most strongly related to 
prey densities that contributed the most to goshawk re- 
production. Competing models were ranked by their ad- 
equacy in explaining the variation in goshawk productiv- 
ity using Akaike Information Criterion (PROC MIXED, 
SAS Institute 1999). To compare the relative importance 
of each prey species, we also used cumulative Akaike 
weights, which were calculated by summing the weights 
across all models that included the variable of interest 
(Burnham and Anderson 2002) . 

Results 

Variation in Goshawk Productivity. The number 
of goshawk territories used to estimate productivity 
was 97 in 1999, 98 in 2000, and 103 in 2001 and 
2002 {N = 401). The proportion of territories with 
active nests was 54% in 1999, 58% in 2000, 28% in 
2001, and 18% in 2002. Goshawk productivity {x ± 
SE) varied among years = 26.78, P< 0.001) 

and ranged from 0.14 ± 0.04 fledglings produced 
per territory in 2002 to 1.23 ± 0.14 fledglings pro- 
duced per territory in 2000 (Fig. 1). There was a 
significant decline = 37.15, P < 0.001) in 

goshawk productivity between 2000 and 2001 (Fig. 

1 ). 

Variation in Prey Density. Ten prey species had 
sufficient detections to estimate density: American 
Robin ( Turdus migratorius) , Clark’s Nutcracker (Nu- 
dfraga Columbiana), Downy Woodpecker (Picoides 
pubescens), golden-man tied ground squirrel (Sper- 
mophilus lateralis). Hairy Woodpecker {Picoides vil- 


losus), Kaibab squirrel {Sciurus aberti kaibabensis) , 
Northern Flicker {Colaptes auratus), red squirrel 
{Tamiasdurus hudsonicus), Steller’s Jay {Cyanodtta 
stelleri), and Williamson’s Sapsucker (Sphyrapicus 
thyroideus). We were unable to estimate densities 
for black-tailed jackrabbit {Lepus califomicus) , Blue 
Grouse {Dendragapus obscurus), chipmunk {Euta- 
mias spp.), cottontail rabbit (Sylvilagus spp.), and 
rock squirrel (Spermophilus variegatus) due to low 
numbers of detections. Detection probability plots 
showed little evidence of heaping, measurement 
errors, and evasive movement prior to detection. 
Total prey density (±SE) varied annually and 
ranged from 2.22 ± 0.08 individuals ha“^ in 2001 
to 3.96 ± 0.14 individuals ha“^ in 2000 (z = 10.39, 
P < 0.001). Density also varied significantly among 
years for most individual prey species (Table 1) in- 
cluding golden-mantled ground squirrel (z = 2.18, 
P — 0.015), Hairy Woodpecker (z = —2.88, P = 
0.002), Kaibab squirrel (z = 2.47, P = 0.007), 
Northern Flicker (z = 5.70, P < 0.001), red squir- 
rel (z = 8.32, P < 0.001), Steller’s Jay (z = 3.25, P 
< 0.001), and Williamson’s Sapsucker (z = —2.78, 
P = 0.003). Significant declines in prey densities 
were also observed between 2000 and 2001 for 
golden-mantled ground squirrel (z = 2.18, P = 
0.015), Kaibab squirrel (z = 2.47, P = 0.007), 
Northern Flicker (z = 2.62, P = 0.005), and red 
squirrel (z = 8.32, P< 0.001), but only red squirrel 


September 2005 


Biology 


241 


Table 1. Annual estimates of Northern Goshawk prey density ha~^ for American Robin (AMRO), Clark’s Nutcracker 
(CLNU), Downy Woodpecker (DOWO), golden-mantled ground squirrel (GMSQ), Hairy Woodpecker (HAWO), 
Kaibab squirrel (KASQ), Northern Flicker (NOFL), red squirrel (RESQ), Steller’sjay (STJA), Williamson’s Sapsucker 
(WISA), and all 10 prey species’ densities combined (Total) on the Kaibab Plateau, Arizona, 1999-2002. 


Species 

1999 

2000 

2001 

2002 

X 

SE 

X 

SE 

X 

SE 

X 

SE 

AMRO 

0.23 

0.05 

0.22 

0.07 

0.27 

0.06 

0.25 

0.06 

CLNU 

0.05 

0.02 

0.05 

0.02 

0.04 

0.01 

0.11 

0.04 

DOWO 

0.20 

0.05 

0.11 

0.03 

0.20 

0.06 

0.11 

0.03 

GMSQ 

0.28 

0.30 

0.64 

0.18 

0.22 

0.06 

0.32 

0.10 

HAWO 

0.09 

0.03 

0.05 

0.02 

0.23 

0.06 

0.17 

0.04 

KASQ 

0.11 

0.04 

0.26 

0.07 

0.07 

0.03 

0.08 

0.02 

NOFL 

0.58 

0.08 

0.77 

0.09 

0.48 

0.06 

0.20 

0.04 

RESQ 

1.16 

0.17 

1.38 

0.15 

0.12 

0.04 

0.23 

0.05 

STJA 

0.41 

0.07 

0.12 

0.05 

0.30 

0.07 

0.33 

0.05 

WISA 

0.18 

0.04 

0.36 

0.08 

0.28 

0.06 

0.45 

0.09 

Total 

3.29 

0.19 

3.96 

0.14 

2.22 

0.08 

2.24 

0.09 


density decreased by an order of magnitude (Table 

1 ). 

Prey density also varied by sampling period (Fig. 
2). However, there were too few observations to 
accurately estimate density by sampling period for 
most individual prey species, so we report only to- 
tal prey density by sampling period. Total prey den- 
sity in the spring sampling period was highest in 
2000, followed by 1999, 2002, and 2001 (Fig. 2). 
However, the decrease in density was only statisti- 


cally significant between 1999 and 2002 (z = 1.74, 
P = 0.041), and 2002 and 2001 (z = 6.58, P = 
0.005). The only significant decrease in total prey 
density between the late-summer sampling period 
of one year and the spring sampling period of the 
next occurred between 2000 and 2001 (z = 6.58, 
P< 0.001; Fig. 2). 

Goshawk Diets. Goshawks on the Kaibab Plateau 
captured and consumed a wide diversity of prey. A 
total of 710 individual prey items consisting of 30 



Figure 2. Total Northern Goshawk prey density estimates ha~^ (±SE) by sampling period on the Kaibab Plateau, 
Arizona, 1999-2002. 


242 


Salafsky et al. 


VoL. 39, No. 3 


Table 2. Prey species each contributing ^3% of all items (N = 710) to Northern Goshawk diets in terms of percent 
frequency and biomass (kg) , and their relationship to the number of fledglings produced per goshawk territory on 
the Kaibab Plateau, Arizona, during 1999—2002. 


Species 

No. 

Percent 

Frequency 


P 

Percent 

Biomass 


P 

Black-tailed jackrabbit 

23 

3 

-0.32 

0.43 

24 

-0.36 

0.40 

Clark’s Nutcracker 

34 

5 

0.03 

0.82 

2 

0.01 

0.88 

Cottontail rabbit 

125 

18 

-0.13 

0.64 

42 

-0.03 

0.84 

Kaibab squirrel 

40 

6 

-0.05 

0.78 

13 

0.02 

0.87 

Northern Flicker 

141 

20 

0.87 

0.07 

8 

0.87 

0.07 

Red squirrel 

87 

12 

0.97 

0.01 

7 

0.95 

0.02 

Steller’s Jay 

88 

12 

-0.09 

0.70 

4 

0.04 

0.81 


species were collected from nest areas during 
1999-2002. Seven species each contributed ^3% 
of all prey items collected in terms of percent fre- 
quency (Table 2) . In descending order of percent 
of total diet, the most common prey items were 
Northern Flickers, cottontail rabbits, red squirrels, 
Steller’s Jays, Kaibab squirrels, Clark’s Nutcrackers, 
and black-tailed jackrabbits. The descending order 
of species biomass contribution to goshawk diets 
was: cottontail rabbits, black-tailed jackrabbits, Kai- 
bab squirrels, Northern Flickers, red squirrels, 
Steller’s Jays, and Clark’s Nutcrackers (Table 2). 
The mean number of prey items per fledgling was 
1.8 in 1999, 2.3 in 2000, 5.1 in 2001, and 7.6 in 
2002. In contrast the mean biomass of prey items 


per fledgling was 0.8 kg in 1999, 0.6 kg in 2000, 
2.3 kg in 2001, and 2.8 kg in 2002. 

Goshawk Productivity and Prey Resources. We 
found a strong positive relationship (r^ = 0.98, P 
= 0.012) between total prey density and goshawk 
productivity from 1999-2002 (Fig. 3) . Although an- 
nual goshawk productivity was highly correlated 
with prey density in the spring sampling period (r^ 
= 0.70, P = 0.163), summer sampling period (r^ 
= 0.75, P = 0.131), and late-summer sampling pe- 
riod (r^ = 0.79, P = 0.112), annual prey density 
accounted for more of the variation in goshawk 
productivity. Based on regression models for each 
prey species, only red squirrel density had a signif- 
icant and positive relationship to goshawk produc- 



Figure 3. The relationship between total prey density ha ^ and the mean number of Northern Goshawk fledglings 
produced per territory on the Kaibab Plateau, Arizona, 1999-2002. 


September 2005 


Biology 


243 


Table 3. Top 10 models for mean number of Northern Goshawk fledglings produced per territory on the Kaibab 
Plateau, Arizona, 1999-2002. Models are ranked based on Akaike’s Information Criteria (AIC) and include model 
covariates, number of parameters (K) , AIC differences (AAIC) and Akaike weights (wj) . 


Model 

AIC 

K 

AAIC 

Wi 

Total prey species 

1134.80 

3 

0.00 

0.70 

Red squirrel 

1137.70 

3 

2.90 

0.16 

Mammal prey species 

1138.10 

3 

3.30 

0.13 

Northern Flicker 

1144.20 

3 

9.40 

0.01 

Total prey in late-summer 

1149.50 

3 

14.70 

0.00 

Hairy Woodpecker 

1150.80 

3 

16.00 

0.00 

Total prey in summer 

1152.20 

3 

17.40 

0.00 

Kaibab squirrel 

1154.20 

3 

19.40 

0.00 

American Robin 

1155.40 

3 

20.60 

0.00 

Total prey in spring 

1156.90 

3 

22.10 

0.00 


tivity (r^ = 0.94, P = 0.031). Red squirrel was also 
the only species that had a significant and positive 
relationship to goshawk productivity for percent of 
diet {P- = 0.97, P — 0.014) and biomass (r^ = 0.95, 
P = 0.024; Table 2). The densities of mammal prey 
species {P = 0.94, P — 0.033) explained more of 
the variation in goshawk productivity than avian 
prey species < 0.01, P = 0.949). 

Our model selection results showed that total 
prey density was clearly the top model (Table 3) . 
This model, which included an annual summation 
of all prey species’ densities, received >70% of the 
Akaike weight across the model set (Table 3) and 
was more than four times as likely as the next best 
model. The only single species models with some 
weight of evidence included those for red squirrel 
and Northern Flicker (Table 3). However, the red 
squirrel density covariate had a higher cumulative 
Akaike weight (99%) than Northern Flicker 
(71%). All other models based on individual prey 
species, avian density, and models of total prey den- 
sity by sampling period had minimal support and 
failed to explain variation in goshawk productivity 
(Table 3). When we compared only the models 
with total prey density by sampling period in a sep- 
arate analysis, total prey density summed over all 
sampling periods was selected as the best model 
(AIC = 811.10, K = 3, AAIC = 0.00, Wj = 0.93). 
All other models, including the model with the dif- 
ference in prey density between late-summer and 
the successive spring (AIC = 873.00, K = 3, AAIC 
= 61.90, Wj = 0.00) and the lowest ranked model 
with late-summer prey density from the prior year 
(AIC = 883.60, K = 3, AAIC = 72.50, Wi = 0.00), 
were not supported by the data. 


Discussion 

A short-term observational study cannot provide 
a strong basis for estimating the causal relationship 
between prey resources and annual goshawk pro- 
ductivity. Thus, our study only established a strong 
association between variation in prey resources 
within the study area and goshawk productivity. Be- 
cause fluctuations in other limiting factors (e.g., 
climate) may have coincided with changes in prey 
resources, we cannot identify the factors ultimately 
responsible for variation in goshawk productivity. 
However, if the patterns we observed between prey 
resources and goshawk productivity were support- 
ed by experimental studies that established a rela- 
tionship between food-supply and goshawk repro- 
duction, then it would be reasonable to infer that 
prey resources may be an important limiting factor 
of goshawk reproduction on the Kaibab Plateau. 

During 1999-2002 we observed high temporal 
correlations between goshawk productivity and an- 
nual prey density; changes in goshawk productivity 
paralleled changes in prey density. Total prey den- 
sity, in addition to the proportion of active gos- 
hawk nests and mean number of fledglings pro- 
duced, was high in 1999 and 2000 and low in 2001 
and 2002. Therefore, it appears that goshawk re- 
production on the Kaibab Plateau responded to 
inter-annual increases in prey density. Several oth- 
er studies have also found close ties between mea- 
sures of goshawk reproduction and the relative 
abundances of prey (Huhtala and Sulkava 1981, 
Doyle and Smith 1994, Keane 1999). Further, gos- 
hawk studies that experimentally manipulated 
food-supply found supplemental food may have in- 


244 


Salafsky et al. 


VoL. 39, No. 3 


fluenced goshawk productivity by increasing nest- 
ling survival when background prey-levels were low 
(Ward and Kennedy 1996, Dewey and Kennedy 

2001) . Thus, we suggest that the number of gos- 
hawk fledglings produced may be influenced by 
fluctuations in prey density. 

On an annual basis, the reproductive responses 
of goshawks depend on the abundance of prey dur- 
ing critical time periods. Low food resources may 
manifest through failure to lay eggs, smaller clutch- 
es, and reduced survival of young (Newton 1998). 
The abundance of prey may be an important de- 
terminant of the “decision” to breed. Goshawks 
initiate breeding before most prey species repro- 
duce, so the density of prey during the incubation 
period is likely similar to prey levels prior to egg- 
laying. On the Kaibab Plateau, prey densities dur- 
ing the incubation stage were similar in 1999 and 
2002, yet goshawk productivity was six times higher 
in 1999. In addition, although there was a signifi- 
cant increase in prey density during the incubation 
period between 2001 and 2002, goshawk produc- 
tivity changed little between these years, suggesting 
that below a density of ca. 0.8 prey ha“^, fewer 
fledglings are produced. However, prey density lev- 
els prior to egg-laying may alter the threshold ef- 
fects of prey density on goshawk productivity 
through physiological constraints. Assuming our 
density estimates represented true population size, 
the difference in prey density between late-summer 
and the next spring should reflect prey density lev- 
els prior to egg-laying. The large decline we ob- 
served in prey density between August 2000 and 
May 2001 indicated that there was substantial over- 
winter mortality for prey species. The lower prey 
numbers prior to egg-laying may have affected the 
ability of females to accumulate sufficient reserves 
to produce eggs in 2001. 

Our results suggest there is a difference in the 
contribution of individual prey species to goshawk 
reproduction. Red squirrel density and their per- 
cent frequency and biomass contribution to gos- 
hawk diet accounted for more variation in goshawk 
productivity than any other species. Although rab- 
bits contributed the majority of biomass to gos- 
hawk diets (>66%), goshawk reproduction was 
lower in most years when rabbits contributed the 
greatest proportion of biomass to the diets. Fur- 
ther, in “poor” goshawk reproductive years (2001, 

2002) , the number of prey items and total biomass 
per fledgling was twice as high as in “good” repro- 
ductive years (1999, 2000). The difference in the 


apparent influence of individual prey species is 
likely a result of encounter rates with goshawks. 
Goshawks are opportunists and will presumably at- 
tempt to capture whatever prey species are readily 
available. However, the limited distributions or dif- 
ferent activity patterns of some prey species de- 
creases the probability that diurnal goshawks will 
encounter them while foraging. Jackrabbits are less 
common in upper elevation forests, and although 
cottontails are widely distributed across the study 
area, they are crepuscular (Hoffmeister 1986). In 
contrast, red squirrels are among the heaviest of 
the diurnal prey species, with a wide distribution 
across the study area (Salafsky 2004) . Red squirrels 
do not hibernate, which likely increases their im- 
portance to goshawks, particularly prior to egg-lay- 
ing. However, the importance of other prey species 
may vary with the spatial distribution of goshawk 
territories relative to the spatial distribution of prey 
habitats. For example, goshawks with territories lo- 
cated primarily within lower elevation forests may 
rely more heavily upon jackrabbits. 

In our study, goshawk productivity on the Kaibab 
Plateau was more closely associated with variation 
in mammal density than in avian density. Goshawks 
may consume more mammals than birds in some 
areas due to the availability and sizes of local prey 
species (Zachel 1985, Widen 1987, Doyle and 
Smith 1994). Similar to our study, Boal and Man- 
nan (1994) and Reynolds et al. (1994) found that 
goshawks on the Kaibab Plateau consumed a high- 
er proportion of mammalian prey. Other goshawk 
studies conducted in northern latitudes identified 
a strong link between goshawk reproductive rates 
and cyclical variation in hare abundance (Mc- 
Gowan 1975, Doyle and Smith 2001). Because an- 
nual variations in predator reproductive rates are 
greatest among species with limited diets that are 
dominated by cyclic prey (Newton 1979), goshawks 
on the Kaibab Plateau may be subject to more 
marked variations in productivity due to their re- 
liance on prey species with fluctuating densities. 

Goshawks have the ability to switch to alternate 
prey when the densities of essential prey species 
are reduced (Doyle and Smith 1994). However, if 
different prey species’ populations decline simul- 
taneously, then the opportunities for goshawks to 
switch to alternative prey species are limited. The 
densities of golden-mantled ground squirrels, Kai- 
bab squirrels, Northern Flickers, and red squirrels 
declined significantly between 2000 and 2001. Fur- 
ther, these species contributed >39% of all prey 


September 2005 


Biology 


245 


items to goshawk diets. Parallel fluctuations in an- 
nual densities of important prey species may result 
in potentially “poor” and “good” years of prey re- 
sources. Thus, it may be that the collective density 
of the entire prey community influences the mag- 
nitude of variation in goshawk productivity on the 
Kaibab Plateau. 

In summary, our results indicate that prey den- 
sity is an important limiting factor of goshawk pro- 
ductivity. Although the temporal correlations be- 
tween goshawk productivity and prey resources 
were consistent over time, other factors may have 
varied with prey density and limited goshawk re- 
production in our study. Synchronous declines in 
prey species’ densities suggests that landscape-level 
factors acting at broad spatial scales, such as cli- 
mate, may interact with prey abundance to limit 
goshawk productivity. Because unfavorable weather 
conditions may have a greater effect on goshawk 
productivity when prey resources are already low, 
it is important to study the relationship between 
goshawk productivity and prey density over long 
time periods and variable environmental condi- 
tions. 

Acknowledgments 

Many thanks go to the goshawk technicians that col- 
lected data for this project. We also thank Clint W. Boal, 
Curtis H. Flather, John J. Keane, Rudy M. King, Marc A. 
Snyder, J. David Wiens, and three anonymous referees for 
reviewing this manuscript. This project was made possible 
with funding provided by both the Rocky Mountain Re- 
search Station and the Southwestern Region of the U.S. 
Department of Agriculture Forest Service. 

Literature Cited 

Boal, C.W. and R.W. Mannan. 1994. Northern Goshawk 
diets in ponderosa pine forests on the Kaibab Plateau. 
Stud. Avian Biol. 16:97-102. 

Buckland, S.T., D.R. Anderson, K.P. Burnham, and J.L. 
Laake. 1993. Distance sampling: estimating abun- 
dance of biological populations. Chapman & Hall, 
London, U.K. 

Burnham, K.P. and D.R, Anderson. 2002. Model selec- 
tion and inference: a practical information-theoretic 
approach, 2nd Ed. Springer-Verlag, New York, NY 
U.S.A. 

Dewey, S.R. and P.L. Kennedy. 2001. Effects of supple- 
mental food on parental-care strategies and juvenile 
survival of Northern Goshawks. Auk 118:352-365. 
Doyle, F.l. and J.M.N. Smith. 1994. Population responses 
of Northern Goshawks to the 10-year cycle in num- 
bers of snowshoe hares. Stud. Avian Biol. 16:122-129. 

AND J.M.N. Smith. 2001. Raptors and scavengers. 

Pages 378—404 in C.J. Krebs, S. Boutin, and R. Boon- 


stra (Eds.), Ecosystem dynamics of the boreal forest 
Oxford University Press, New York, NYU.S.A. 

Dunning, J.B., Jr. 1993. GRC handbook of avian body 
masses. CRC Press, Inc., Boca Raton, FL U.S.A. 

Emlen, J.T. 1971. Population densities of birds derived 
from transect counts. Auk 88:332-334. 

Gotelli, N.J. 1998. A primer of ecology. Second edition. 
Sinauer Associates, Inc., Sunderland, MA U.S.A. 

Hoffmeister, D.F. 1986. Mammals of Arizona. The Uni- 
versity of Arizona Press, Tucson, AZ U.S.A. 

Huhtala, K. and S. Sulkava. 1981. Environmental influ- 
ences on goshawk breeding in Finland. Pages 89-104 
in R.E. Kenward and I.M. Lindsay (Eds.), Understand- 
ing the goshawk. The International Association for 
Falconry and Conservation of Birds of Prey, Oxford, 
England. 

Joy, S.M. 2002. Northern Goshawk habitat on the Kaibab 
National Forest in Arizona: factors affecting nest lo- 
cations and territory quality. Ph.D. dissertation, Col- 
orado State Univ., Fort Collins, CO U.S.A. 

Keane, JJ- 1999. Ecology of Northern Goshawk in the 
Sierra Nevada, California. Ph.D. dissertation, Univ. of 
California, Davis, CA U.S.A. 

Kennedy, P.L. 1991. Reproductive strategies of Northern 
Goshawks and Cooper’s Hawks in north-central New 
Mexico. Ph.D. dissertation, Utah State Univ., Logan, 
UT U.S.A. 

Marti, C.D. 1987. Raptor food habitat studies. Pages 67- 
80 in B.G. Pendleton, B.A. Millsap, K.W. Cline, and 
D.M. Bird, [Eds.], Raptor management techniques 
manual. National Wildlife Federation, Sci. Tech. Se- 
ries No. 10, Washington, DC U.S.A. 

Martin, T.E. 1987. Food as a limit on breeding birds: a 
life-history perspective. Ann. Rev. Ecol. Syst. 18:453- 
487. 

McGowan, J.D. 1975. Distribution, density and produc- 
tivity of goshawks in interior Alaska. Final Report to 
Federal Aid in Wildlife Restoration, Project Report 
W17-3, 4, 5, 6, Alaska Department of Fish and Game, 
Juneau, AK U.S. A. 

Newton, I. 1979. Population ecology of raptors. Buteo 
Books, Vermillion, SD U.S. A. 

, 1998. Population limitation in birds. Academic 

Press, Inc., San Diego, CA U.S.A. 

Reynolds, R.T., R.T. Graham, M.H. Reiser, R.L. Bassett, 
P.L. Kennedy, D.A. Boyce, Jr., G. Goodwin, R. Smith, 
AND E.L. Fisher. 1992. Management recommenda- 
tions for the Northern Goshawk in the southwestern 
United States. Gen. Tech. Report RM-217. USDA For- 
est Service, Rocky Mountain Research Station, Fort 
Gollins, CO U.S.A. 

AND S.M. Joy. 2005. Demography of Northern 

Goshawks in northern Arizona, 1991-1996. Stud. Avi- 
an Biol.', in press. 

, S.M. Joy, and D.G. Leslie. 1994. Nest productiv- 
ity, fidelity, and spacing of Northern Goshawks m 
northern Arizona. Stud. Avian Biol. 16:106-113. 


246 


Salafsky et al. 


VoL. 39, No. 3 


AND E.C. Meslow. 1984. Partitioning of food and 

niche characteristics of coexisting Accipiter during 
breeding. Auk 101:761-779. 

, G.C. White, S.M. Joy, and R.W. Mannen. 2004. 

Effects of radiotransmitters on Northern Goshawks: 
do tailmounts lower survival of breeding males? J. 
Wildl. Manage. 68:25—32. 

, J.D. Wiens, S.M. Joy, and S.R. Salafsky. 2005. 

Sampling considerations for demographic and habitat 
studies of Northern Goshawks. J. Raptor Res. 39:274— 
285. 

, J.D. Wiens, and S.R. Salafsky. 2005. A review and 

evaluation of factors limiting Northern Goshawk pop- 
ulations. Stud. Avian Biol.: in press. 

Salafsky, S.R. 2004. Covariation between prey abundance 
and Northern Goshawk fecundity on the Kaibab Pla- 
teau, Arizona. M.S. thesis, Colorado State Univ., Fort 
Collins, CO U.S.A. 

SAS Institute. 1999. SAS/STAT User’s Guide, Version 8. 
SAS Institute, Inc., Cary, NC U.S.A. 

Squires, J.R. and R.T. Reynolds. 1997. Northern Gos- 
hawk {Accipiter gentilis) . Pages 1-32 in A. Poole and F. 


Gill [Eds.], The Birds of North America, No. 298. The 
Academy of Natural Sciences, Philadelphia, PA U.S.A. 

Thomas, L., J.L. Laake, J.F. Derry, S.T. Buckland, D.L. 
Borchers, D.R. Anderson, K.P. Burnham, S. Strind- 
berg, S.L. Hedley, M.L. Burt, F. Marques, J.H. Pol- 
lard, AND R.M. Fewster. 1998. Distance 3.5. Research 
Unit for Wildlife Population Assessment, Univ. of St. 
Andrews, U.K. 

Ward, J.M. and P.L. Kennedy. 1996. Effects of supple- 
mental food on growth and survival of juvenile North- 
ern Goshawks. Auk 113:200—208. 

Widen, P. 1987. Goshawk predation during winter, spring, 
and summer in a boreal forest in central Sweden. Hol- 
arctic Ecol. 10:104-109. 

Wiens, J.D. and R.T. Reynolds. 2005. Is fledgling success 
a reliable index of fitness in Northern Goshawks? J. 
Raptor Res. 39:210—221. 

Zachel, C.R. 1985. Food habits, hunting activity, and 
post-fledgling behavior of Northern Goshawks {Accip- 
iter gentilis) in interior Alaska. M.S. thesis, Univ. Alas- 
ka, Fairbanks, AK U.S.A. 

Received 3 March 2004; accepted 15 May 2005 

Associate Editor: Clint Boal 


J. Raptor Res. 39(3):247-252 
© 2005 The Raptor Research Foundation, Inc. 

A SKEWED SEX RATIO IN NORTHERN GOSHAWKS: 

IS IT A SIGN OF A STRESSED POPULATION? 

Michael F. Ingraldi^ 

Arizona Game and Fish Department, Research Branch, 2221 W. Greenway Road, 

Phoenix, AZ 85023 US. A. 

Abstract. — I examined 6 yr (1993-98) of data on Northern Goshawk {Accipiter gentilis) reproductive 
performance in east-central Arizona. Forty-four breeding territories were monitored over the 6-yr period, 
yielding 109 nesting attempts and 141 fledglings produced from 76 successful broods. Among the 63 
broods from which sex of nestlings could be determined by measurements, 29% fledged one young, 
54% fledged two young, and 17% fledged three young. The mean sex ratio across years was 1.93 ± 
0.70 (±95% Cl) males/females (annual range 1.1:1-3:1). When combining all fledglings of known sex 
(76 males and 43 females), a 1.77:1 male to female sex ratio was significantly different from 1:1 (P = 
0.002). A male-biased fledgling sex ratio may be explained by one or more of the following hypotheses: 
(1) more males were produced due to nutritional stress resulting in minimization of investment in the 
larger sex (females) and (2) fewer females were produced because of differential mortality due to 
exposure to the elements during the nestling stage. I propose that environmental stress in the form of 
exposure to the elements during a critical life stage (e.g., rainfall during the nestling stage), combined 
with limited food availability, may be driving the skewed sex ratios observed in this local Northern 
Goshawk population. 

Key Words: Northern Goshawk; Accipiter gentilisr, nestling, sex ratio; sex allocation. 


iES EL COCIENTE DE SEXOS SESGADO DE ACCIPITER GENTILIS UNA SENAL DE UNA POBLA- 
CION ESTRESADA? 

Resumen. — En este estudio examine datos sobre el desempeno reproductivo de Accipiter gentilis reco- 
lectados a lo largo de seis ahos (1993-98) en el centro-este de Arizona. Un total de 44 territorios de 
nidificacion fueron monitoreados a traves de los seis anos, observando un total de 109 intentos de 
nidificacion y 141 volantones producidos en 76 nidadas exitosas. De las 63 nidadas en las que el sexo 
de los pichones pudo ser determinado mediante mediciones, el 29% produjeron un volanton, el 54% 
dos volantones y el 17% tres volantones. El cociente de sexos promedio a traves de los anos fue de 1.93 
± 0.70 (±95% IC) machos/hembras (rango anual 1.1:1-3:1). A1 combinar todos los volantones de sexo 
conocido (76 machos y 43 hembras), el cociente de machos a hembras resultante de 1.77:1 fue signi- 
ficativamente diferente de 1:1 (P = 0.002). Un cociente de sexos sesgado hacia los machos en los 
volantones podrfa explicarse por las siguientes hipotesis: (1) se produjeron mas machos como conse- 
cuencia de estres nutricional que llevo a minimizar la inversion en el sexo de mayor tamaho (hembras) 
y/o (2) se produjeron menos hembras como resultado de una mortalidad diferencial debida a la ex- 
posicion al ambiente durante la permanencia de los pichones en el nido. Propongo que el estres 
ambiental causado por la exposicion a las condiciones ambientales durante una etapa critica de la vida 
(e.g., la Iluvia durante la etapa de crianza en el nido), combinado con la disponibilidad limitada de 
alimento, podrian estar determinando el sesgo en el cociente de sexos observado en esta poblacion 
local de A. gentilis. 

[Traduccion del equipo editorial] 


Female Northern Goshawks {Accipiter gentilis) are 
approximately 20—30% heavier than males (New- 
ton 1979, Palmer 1988, Squires and Reynolds 
1997). Larger offspring require more food to at- 


^ Email address: mingraldi@cybertrails.com 


tain fledging age, and thus are more susceptible to 
food shortages (Glutton-Brock et al. 1985, Teather 
and Weatherhead 1988, Anderson et al. 1993). An 
underlying assumption of an adjustable sex ratio is 
that parents should bias the sex ratio of their off- 
spring toward the sex whose production will most 
increase their own fitness (Trivers and Willard 


247 


248 


Ingraldi 


VoL. 39, No. 3 


1973, Clark 1978, Charnov 1982, Glutton-Brock 
1986). The sex ratio at fledging should be the in- 
verse of the ratio of the cost of producing each sex 
(Fisher 1930) . Consequently, under times of limit- 
ed prey resources the sex ratio should be biased 
toward the “cheaper sex” (in Northern Goshawks: 
the male; Howe 1977, Cronmiller and Thompson 
1981, Teather and Weatherhead 1988). In sexually- 
dimorphic species, differential survival between 
the sexes is most likely due to differing nutritional 
requirements, with higher mortality incurred by 
the larger sex (Torres and Drummond 1997, Shel- 
don et al. 1998). However, this may be offset in 
Northern Goshawks by the tendency of the female 
nestlings to seize prey items more readily than 
males (Lee 1980). 

Sex allocation in sexually dimorphic raptors has 
provided mixed observations. Studies have shown 
prevalence toward male fledglings in Golden Ea- 
gles (Aquila chrysaetos; Edwards et al. 1988), Har- 
ris’s Hawks (Parabuteo unicinctus; Bednarz and Hay- 
den 1991), Montagu’s Harriers {Circus pygargus; 
Leroux and Bretagnolle 1996), and Cooper’s 
Hawks {Accipiter cooperii; Rosenfield et al. 1996). 
Other studies showed prevalence toward female 
fledglings in Northern Harriers {Circus cyaneus; 
Balfour and Cadbury 1979), Bald Eagles {Halieaee- 
tus leucocephalus; Bortolotti 1986), and Peregrine 
Falcons {Falco peregrinus; Olsen and Cockburn 
1991). However, most studies on raptors have re- 
vealed equal sex ratios in offspring (Newton 1979). 
In some previous work on Northern Goshawks, the 
sex ratio for fledglings has been shown not to dif- 
fer from 1:1 (Newton 1979, Reynolds and Joy 
1998). 

A male-biased fledgling sex ratio may be ex- 
plained by one or more of the following hypothe- 
ses: (1) more males were fledged due to nutritional 
stress resulting in minimization of investment in 
the larger sex (females) during the nestling stage 
(Trivers and Willard 1973, Torres and Drummond 
1997) and (2) fewer females were produced be- 
cause of differential mortality due to exposure to 
the elements during the nestling stage with larger 
young exhibiting a higher mortality rate (Newton 
1979:136-137). Hypothesis 1 above may be a direct 
product of adaptive selection, whereas hypothesis 
2 may simply be a nonadaptive by-product of dif- 
ferential mortality. The purpose of this paper is to 
discuss these potential explanations for the sex ra- 
tio exhibited by Northern Goshawk fledglings in 
east-central Arizona. 


Study Area 

The Sitgreaves portion of the Apache-Sitgreaves Na- 
tional Forest is located on the Mogollon Plateau in east- 
central Arizona and encompasses approximately 330 300 
ha. Elevation ranges from 1768-2417 m. To the south, 
the study area is bounded by the Mogollon Rim, a large 
escarpment extending east across central Arizona and 
into New Mexico. To the east, the study area is bounded 
by the Springerville Ranger District boundary on the 
Apache National Forest. A wide range of vegetation com- 
munities occurs within the study area (Brown 1994) . The 
Mogollon Rim edge is dominated by deep drainages with 
mixed-conifer communities of Douglas-fir {Pseudotsuga 
menziesii), white fir {Abies concolor), and ponderosa pine 
{Pinus ponderosa), with pockets of aspen {Populus tremu- 
hides). New Mexico locust {Robinia neomexicana) , and 
Gambel oak {Quercus gambelii). Ridgetops are generally 
dominated by ponderosa pine forest. To the north, as 
elevation decreases, a ponderosa pine/juniper-pinyon 
ecotone transitions to a juniper-pinyon woodland domi- 
nated by alligator juniper (Juniperus deppeana), Utah ju- 
niper (/. osteosperma) , and Rocky Mountain pinyon pine 
{P. edulis). As elevation decreases further, a plains grass- 
land community develops, dominated by blue grama 
{Bouteloua gracilis), sand dropseed {Sporobolus cryptan- 
drus) , and fourwing saltbush {Atriplex canescens ) . 

Methods 

I monitored demographic parameters of a Northern 
Goshawk population on the Apache-Sitgreaves National 
Forest from 1993-98 and specifically noted the sex of 
fledgling birds. I visited occupied nests periodically dur- 
ing the breeding season (ca. late April-early August) to 
monitor status and productivity. An occupied nest was 
one in which at least one egg was laid (usually inferred 
by observing a bird in incubation posture). I estimated 
the ages of nestlings using a photographic guide pro- 
duced by Boal (1994). Birds were deemed to have sur- 
vived to fledge when they were greater than 80% of fledg- 
ling age (31 d old; Steenhof 1987). When nestlings were 
between 30-40 d old, I banded them with U.S. Geological 
Survey aluminum leg bands and took the following stan- 
dard morphological measurements: tarsus dorsal-ventral 
(smallest front to back measurement of the tarsus to 
nearest 0.1 mm), tarsus lateral (smallest side to side mea- 
surement of the tarsus to nearest 0.1 mm), hallux (tip of 
talon to the beginning of the fleshy pad of the hind toe 
to nearest 0.1 mm), bill depth (perpendicular to the bill 
from the top in front of the cere to the bottom of the 
bill to nearest 0.1 mm), culmen (top portion of the bill 
in front of the cere to the tip of the bill to nearest 0.1 
mm), mass (taken to nearest 10 g). I used a A-means 
cluster analysis (SPSS 1997) to determine if morpholog- 
ical measurements could adequately identify sexes of 
nestlings. Only birds from broods in which all the mea- 
surements were collected were included in the analysis. 
In 1999, I collected blood samples from 22 nestlings and 
had them genetically analyzed for sex identification (Avi- 
an Biotech International, Tallahassee, FL U.S.A.) using a 
polymerase-chain reaction to amplify CHD (chromo-he- 
licase-DNA binding) genes Z and W that are located on 
the avian sex chromosomes in birds. For detailed proce- 
dural information, see Griffiths et al. (1998), Fridolfsson 


September 2005 


Biology 


249 


Table 1. Nestling sex-ratio and productivity of Northern Goshawks from territories monitored on the Apache-Sit- 
greaves National Forest, Arizona, (1993—98). 


Year 

No. Nests 
Monitored 

Nesting 

Occupancy 

Rate=^ 

No. 

Successful 

Broods 

No. Male 
Fledglings 

No. Female 
Fledglings 

No. 

Fledglings 
OF Unknown 
Sex’’ 

Sex 

Ratio‘s 

pd 

1993 

30 

0.69 

17 

10 

9 

11 

1.1 

0.819 

1994 

33 

0.33 

8 

9 

3 

4 

3 

0.083 

1995 

39 

0.66 

14 

17 

10 

0 

1.7 

0.178 

1996 

42 

0.52 

16 

16 

8 

4 

2 

0.102 

1997 

42 

0.31 

6 

7 

3 

0 

2.3 

0.206 

1998 

44 

0.43 

15 

17 

10 

3 

1.7 

0.178 

Total 

230 

X = 0.49 

76 

76 

43 

22 

1.7 

0.003 


“ The number of occupied nests (a nest where at least one egg was laid) /total number of nesting territories monitored. 
Fledglings not measured, therefore gender indeterminate. 

Male/female. 

Significance of Chi-square test for difference from a 1:1 sex ratio. 


and Ellegren (1999), and Avian Biotech International 
(2005) . I calculated the sex ratio as the number of male 
per female fledglings. I used a chi-square analysis to test 
if the sex ratio was significantly different from 1:1 (Zar 
1984). I used Spearman’s correlation analysis to test for 
significance in relationships between demographic pa- 
rameters (e.g., sex ratio versus the number of territories 
with an occupied nest per the number of territories mon- 
itored). All statistical tests were deemed significant at P 
< 0.05, and all means were expressed ±95% Confidence 
Interval (Cl) . I compiled monthly summaries of total pre- 
cipitation from U.S. National Oceanic and Atmospheric 
Administration records collected at the Show Low, Ari- 
zona, municipal airport (elevation = 1950 m) located 
within the study area. 

Results 

Breeding territories were monitored over the 6- 
yr period (range = 30-44 per yr), yielding 109 
nesting attempts and 141 fledglings produced from 
76 successful broods (Table 1). Among the 63 


broods from which gender could be determined, 
29% fledged one young, 54% fledged two young, 
and 17% fledged three young (Table 2). Two rel- 
atively homogenous groups of nestlings were dis- 
cerned using a /5-means cluster analysis, and their 
morphological measurements showed minimal 
overlap (Table 3). The two groups are easily rec- 
ognized when the lateral tarsus and culmen length 
measurements are plotted (Fig. 1). I considered all 
members of the larger group as females. The mea- 
surements used above correctly classified to gender 
all 22 nestlings that were genetically analyzed in 
1999. The mean sex ratio across years was 1.93 ± 
0.70 (±95% Cl) males/female (range = 1. 1:1-3: 
1). When combining all fledglings of known sex 
(76 males and 43 females), a 1.77:1 male to female 
sex ratio resulted that was significantly different 
from 1:1 (x^ — 9.15, df = 1, P = 0.002). 


Table 2. Observed brood size and sex ratio of Northern Goshawks fledged on the Apache-Sitgreaves National Forest, 
Arizona, (1993-98). 


Nesting Outcome 


Brood Size 


1 

2 

3 

All males 

13 

13 

2 

All females 

5 

7 

0 

1 male and 1 female 

— 

14 

— 

2 males and 1 female 

— 

— 

8 

1 male and 2 females 

— 

— 

1 

Sex ratio (M/F) 

13/5 = 2.6 

40/28 = 1.43 

23/10 = 2.3 

P* 

0.06 

0.15 

0.02 


^ Significance of Chi-square test for difference from a 1:1 sex ratio. 


250 


Ingraldi 


VoL. 39, No. 3 


Table 3. Summary statistics of Northern Goshawk morphological measurements taken on the Apache-Sitgreaves 
National Forest, Arizona (1993-98), 


Group 

Measurement (x ± 2 SE, Range) 

a 

Fledgling Group 1 
(Presumed Male) 
AT= 59 

Fledgling Group 2 
(Presumed Female) 
N= 34 

Adult Females 
N= 25 

Culmen (mm) 

19.05 ± 0.27 

21.36 ± 0.34 

24.53 ± 0.34 


16.2-21.7 

18.5-23.4 

23.0-26.3 

Beak depth (mm) 

14.63 ± 0.18 

16.63 ± 0.21 

17.93 ± 0.18 


13.3-16.1 

15.1-17.8 

17.2-18.7 

Tarsus D/V*’ (mm) 

7.04 ± 0.13 

8.38 ± 0.19 

10.48 ± 0.36 


6.0-8. 1 

7.4-9.3 

8.7-12.0 

Tarsus (mm) 

5.36 ± 0.07 

6.50 ± 0.13 

7.48 ± 0.15 


4.8-6.0 

5.9-7.4 

6.9-S.5 

Hallux (mm) 

23.79 ± 0.36 

27,02 ± 0.52 

31.98 ± 0.39 


19.6-27.0 

24.1-31.1 

30.2-33.8 

Mass (g) 

702 ± 14 

894 ± 36 

1026 ± 39 


565-810 

620-1085 

845-1265 


See text for detailed description of measurements, 

D/V = the smallest front to back measurement of the tarsus (dorsal/ventral). 
' L = the smallest side to side measurement of the tarsus (lateral) . 


Discussion 

Hypothesis 1 — Nutritional Stress Would Mini- 
mize the Investment in Larger Sex. Meyers (1978) 
predicted that during periods of lower than aver- 
age resource abundance, offspring sex ratio in a 
population should shift toward the sex having the 
lower energy needs. In times of plentiful resources, 
adults should invest in the larger offspring, which 
benefit from greater size. I observed a negative re- 
lationship between the annual sex ratio and the 



24.0 -1 

'i 

23.0 - 

E 

22.0 - 

£ 

o> 

21.0 - 

c 

V 

20.0 - 

c 

0) 

19.0 - 

E 

3 

18.0 - 

u 

17.0 - 

16.0 - 


4.0 


• \, *X 


m 

••• 




« ♦ 

♦ * 

* 

t 


• ••• 


5.0 6.0 7.0 

Tarsus lateral (mm) 


8.0 


Figure 1. Plot of the culmen length and lateral tarsus 
measurements for Northern Goshawk nestlings mea- 
sured on the Apache-Sitgreaves National Forest, AZ 
(1993-98). The two groups (circles presumed to be 
males and the diamonds females) were distinguished us- 
ing a fe-means cluster analysis of all morphological mea- 
surements collected. 


nesting occupancy rate (Fig. 2A) . A low nesting oc- 
cupancy rate most likely reflects a lower than av- 
erage abundance of some resource such as prey 
availability. Conversely, in years with a high nesting 
occupancy rate, which may have reflected an above 
average resource availability, the number of female 
offspring was greater (Fig. 2B). Yet, for this hy- 
pothesis to be an adaptive selection strategy, I 
would expect to find a sex ratio skewed toward fe- 
males when the nesting occupancy was high. In- 
stead when the nesting activity rate was high the 
sex ratio approached 1:1. This result may lend cre- 
dence to the resource-shortage hypothesis being a 
nonadaptive by-product of differential mortality 
(Weatherhead and Teather 1991). 

Hypothesis 2 — Male Nestlings are Less Suscep- 
tible to Adverse Weather Conditions. Newton 
(1979) suggests that females would be more sus- 
ceptible to mortality from exposure than smaller 
males. When being brooded, smaller nestlings may 
be more sheltered from environmental elements 
than larger siblings; or during episodes of rain, 
larger drenched female nestlings may hold more 
water and require more time to dry. Thus, larger 
female nestlings would remain cooler longer and 
be more susceptible to hypothermia. In years when 
more rainfall occurred during the nestling period 
(May and June), I observed more male fledglings 


September 2005 


Biology 


251 



3.5 n 


3 - 


2.5 - 


o 


CO 

2 - 

DC 


X 

1.5 - 

(D 


CO 

1 J 


0.2 0.4 0.6 0.8 

Nesting Activity Rate 



Fecundity 

O O O O 
3 Kd G) 00 

1 1 1 1 1 

• 

♦ 

♦ ♦ ♦ 

♦ 

u n 

0 “ 

1 \ 1 

2 0.4 0.6 0.8 

Nesting Activity Rate 


Figure 2. (A) Sex ratio of fledgling Northern Goshawks 

plotted against the annual nesting occupancy rate (num- 
ber of occupied nests/total number of nests monitored) 
on the Apache-Sitgreaves National Forest, AZ, 1993-98 (r 
= —0.84, P = 0.04). (B) Fecundity rate (number of fe- 
male fledglings produced/ total number of breeding fe- 
males) plotted against the nesting occupancy rate of 
Northern Goshawks on the Apache-Sitgreaves National 
Forest, AZ, 1993-98 (r = 0.77, P = 0.07). 

produced (Fig. 3) . This phenomenon of induced 
mortality due to exposure may also be compound- 
ed by nutritional stress (i.e., shortage of prey). Wet 
weather conditions have been shown to prevent 
adult raptors from hunting efficiently (Hiraldo et 
al. 1990, Kostrzewa and Kostrzewa 1990). Times of 
food shortage may force the nesting adult female 
to leave the nest in search of prey, thereby increas- 
ing the risk of exposure to the nestlings. For ex- 
ample, Boal et al. (2005) reported the failure of 
several nests in Minnesota after a 10-d period of 
constant rain. 

Conclusion 

I suggest that limited food availability combined 
with environmental stress in the form of exposure 
to the elements during a critical life stage (e.g., 
rainfall during the nestling stage) may be driving 
the skewed sex ratios observed in this local North- 
ern Goshawk population. Of the two hypotheses 



3.5 n 

II 


S 

3 - 

o 

2.5 - 

CO 

DC 

2 - 

X 

0) 

1.5 


1 -- 


0 2 4 6 8 

Combined Rainfall for May and June (cm) 


Figure 3. Relationship between the sex ratio of fledg- 
ling Northern Goshawks monitored from 1993-98 and 
rainfall (r = 0.81, P = 0.05) measured at the U.S. Na- 
tional Oceanic and Atmospheric Administration weather 
station located at the Show Low, Arizona, airport within 
the Apache-Sitgreaves National Forest, during the brood- 
rearing period (May and June) . 

presented, the available evidence supports that 
Northern Goshawks may exhibit selection toward 
the minimization of investment in the larger sex 
(i.e., support for the nutritional-stress hypothesis). 
But at this time the compounding phenomena of 
a potential increase in exposure time of nestlings 
because of the decrease in the adult female nest 
attentiveness due to a possible decrease in prey 
availability cannot be discarded. During the early 
nestling period, monitoring adult female nest at- 
tentiveness (i.e., her sheltering of the nestlings) 
and the amount of prey brought to the nestlings 
could help tease apart these two potential expla- 
nations that may be driving the skewed sex ratio 
(i.e., exposure or food shortage). 

Acknowledgments 

I would like to thank the 1993-98 field crews, especially 
Renee Wilcox, John Koloszar, and Nora Schubert, for 
their hard work and dedication over the years. A special 
gratitude goes to my wife. Merry, for putting up with all 
the field crews, crew living conditions, and time away. 
Thanks to Glenn Proudfoot for assisting in my under- 
standing of avian genetics and genetic analysis. Lastly, I 
appreciate all the time and energy spent by the reviewers 
and editors that worked on this manuscript. 

Literature Cited 

Anderson, D.J.,J. Reeve, J.E.M. Gomez, W.W. Weathers, 
S. Hutson, H.V. Cunningham, and D.M. Bird. 1993. 
Sexual size dimorphism and food requirements of 
nestling birds. Canadian J. Zoology 71:2541-2545. 
Avian Biotech International. 2005. Avian Biotech 
International, Tallahassee, FL U.S.A. <http://www. 
avianbiotech.com/sexing_center.htm. > 

Balfour, E. and CJ. Cadbury. 1979. Polygyny, spacing, 
and sex ratio among Hen Harriers Circus cyaneus in 
Orkney, Scotiand. Ornis Scandinavica 10:133—141. 


252 


Ingraldi 


VoL. 39, No. 3 


Bednarz, J.C. AND TJ. Hayden. 1991. Skewed brood sex 
ratio and sex biased hatching sequence in Harris’s 
Hawks. Am. Nat. 137:116-132. 

Boal, C. 1994. A photographic and behavioral guide to 
aging nestling Northern Goshawks. Stud. Avian Biol. 
16:32-40. 

AND R.W. Mannan. 1994. Northern Goshawk diets 

in ponderosa pine forests on the Kaibab Plateau. Stud. 
Avian Biol. 16:97-102, 

AND D.E. Andersen, and P.L. Kennedy. 2005. Pro- 
ductivity and mortality of Northern Goshawks in Min- 
nesota. J. Raptor Res. 39:222-228. 

Bortolotti, G.R. 1986. Influence of sibling competition 
on nestling sex ratios of sexually dimorphic birds. Am. 
Nat. 127:495-507. 

Brown, D.E. 1994. Biotic communities: southwestern 
United States and northwestern Mexico. Univ. Utah 
Press, Salt Lake City, UT U.S.A. 

Charnov, E.L. 1982. The theory of sex allocation. Prince- 
ton Univ. Press, Princeton, NJ U.S.A. 

Clark, A.B. 1978. Sex ratio and local resource competi- 
tion in a prosimian primate. Science 201 :1&5-165. 

Clutton-Brock, T.H. 1986. Sex ratio variation in birds. 
Ibis 128:317-329. 

, S. Albon, and re. Guinness. 1985. Parental in- 
vestment and sex differences in juvenile mortality in 
birds and mammals. 313:131-133. 

Cronmiller, J.R. and C.R. Thompson. 1981. Sex ratio 
adjustment in malnourished Red-winged Blackbird 
broods. J. Field Ornith. 52:65-67. 

Edwards, T.C., Jr., M.W. Collopy, K. Steenhof, and 
M.N. Kochert. 1988. Sex ratios of fledgling Golden 
Eagles. Auk 105:793-796, 

Fisher, R.A. 1930. The genetical theory of natural selec- 
tion. Oxford Univ. Press, Oxford, U.K. 

Fridolfsson, A.K. and H. Ellegren. 1999. A simple and 
universal method for molecular sexing of non-ratite 
birds./. Avian Bio. 30:116-121. 

Griffiths, R., M.C. Double, K. Orr, and RJ .G. Dawson. 
1998. A DNA test to sex most birds. Molecular Ecology 
7: 1071-1075. 

Hiraldo, R, J.R Veiga, and M. Manez. 1990. Growth of 
nestling Black Kites Milvus migrant, effects of hatching 
order, weather and season. T. Zool. Soc. Lond. 222:197- 
214. 

Howe, H.R 1977. Sex-ratio adjustment in the Common 
Grackle. Science 198:744—746. 

Kostrzewa, a. and R. Kostrzewa. 1990. The relationship 
of spring and summer weather with density and 
breeding performance of the buzzard Buteo buteo, gos- 
hawk Accipiter gentilis and kestrel Falco tinnunculus. Ibis 
132:550-559. 

Lee, J.A. 1980. Survival of the smallest nestling in gos- 
hawks. /. Raptor Res. 14:70-72. 


Leroux, a. and V. Bretagnolle. 1996. Sex ratio varia- 
tions in broods of Montagu’s Harriers Circus pygargus 
J. Avian Biol. 27:63-69. 

Meyers, J.H. 1978. Sex ratio adjustments under food 
stress: maximization of quality or numbers of off- 
spring? Am. Nat. 112:381-388. 

Newton, I. 1979. Population ecology of raptors. Buteo 
Books, Vermillion, SD U.S.A. 

Olsen, P.D. and A. Cockburn. 1991. Female-biased sex 
allocation in Peregrine Falcons and other raptors. Be- 
hav. Ecol. Sociobiol. 28:417-423. 

Palmer, R.S. 1988. Handbook of North American birds. 
Vol. 4. Diurnal raptors. Yale University Press, New Ha- 
ven, CT U.S.A. 

Reynolds, R.T. and S.M. Joy. 1998. Distribution, territory 
occupancy, dispersal, and demography of Northern 
Goshawks on the Kaibab plateau, Arizona. Final Rep. 
Heritage Project No, 194045, Arizona Game and Fish 
Dept., Phoenix, AZ U.S.A. 

Rosenfield, R.N., J. Bielefeldt, and S.M. Vos. 1996. 
Skewed sex ratios in Cooper’s Hawk offspring. Auk 
113:957-960. 

Sheldon, B.C., J. Merila, G. Lindgren, and H. Elle- 
gren. 1998. Gender and environment sensitivity in 
nestling Collared Flycatchers. Ecology 79:1939-1948. 

Squires, J.R. and R.T. Reynolds. 1997. Northern Gos- 
hawk {Accipiter gentilis) . In A. Poole and F. Gill [Eds.], 
The Birds of North America, No. 298. The Academy 
of Natural Sciences, Philadelphia, PA and The Amer- 
ican Ornithologists’ Union, Washington, DC U.S.A. 

SPSS. 1997. User’s guide. Base 7.5 for Windows, SPSS 
Inc., Chicago, IL U.S.A. 

Steenhof, K. 1987. Assessing raptor reproductive success 
and productivity. Pages 157-170 in B.A. Pendleton, 
B.A. Millsap, K.W. Kline, and D.M. Bird [Eds.], Rap- 
tor management techniques manual. Natl. Wildl. Fed. 
Sci. Tech. Ser. No. 10., Washington, DC U.S.A. 

Teather, K.L. and P.J. Weatherhead. 1988. Sex-specific 
energy requirements of Great-tailed Grackle (Quisca- 
lus mexicanus) nestlings. J. Anim. Ecol. 57:659 — 668. 

Torres, R. and H. Drummond. 1997. Female-biased mor- 
tality in nestlings of birds with size dimorphism. J. 
Anim. Ecol. 66:859-865. 

Trtvers, R.L. and D.E. Willard. 1973. Natural selection 
of parental ability to vary the sex ratio of offspring. 
Science 179:90—92. 

Weatherhead, PJ- AND K.L. Teather. 1991. Are skewed 
fledgling sex ratios in sexually dimorphic birds adap- 
tive? Am. Nat. 138:1159-1172. 

Zar, J.H. 1984. Biostatistical analysis. 2nd edition. Pren- 
tice-Hall, Inc., Englewood Cliffs, NJ U.S.A. 

Received 2 February 2004; accepted 9 March 2005 

Associate Editor: Michael I. Goldstein 


J Raptor Res. 39(3) :253— 263 
© 2005 The Raptor Research Foundation, Inc. 


NORTHERN GOSHAWK {ACCIPITER GENTILIS LAINGI) POST- 
FLEDGING AREAS ON VANCOUVER ISLAND, BRITISH COLUMBIA 

Erica L. McClaren^ 

BC Ministry of Environment, 2080 Labieux Road, Nanaimo, BC V9T 6J9 Canada 

Patricia L. Kennedy 

Eastern Oregon Agricultural Research Center, Oregon State University, RO. Box E, Union, OR 97883 US. A. 

Donald D. Doyle 

BC Ministry of Environment, 2080 Labieux Road, Nanaimo, BC V9T 6J9 Canada 

Abstract. — The area used by immature birds from the time they fledge until independence is the post- 
fledging area (PEA). Published estimates of PFA size (170 ha) are only available from a Northern 
Goshawk {Accipiter gentilis atricapillus) population in New Mexico and applicability of this estimate to 
other regions and habitat types is unknown. Our objectives were to estimate PFA size and length of the 
post-fledging period for Northern Goshawk (A. g. laingi) nests on Vancouver Island, British Columbia. 

We estimated PFA size from 95% adaptive kernel estimates of telemetry locations from 12 fledglings at 
12 nests between 29 June and 2 September 2001-02 (N = 6, 2001; N = 6, 2002). Because our adaptive 
kernel estimates are based on a small number of locations, we also estimated the precision of these 
home range estimates using a smoothed bootstrap approach. Almost all (93%) fledgling locations were 
within 200 m of nests during the early fledgling-dependency period, but less than half (42.4%) of these 
locations were within this distance during the late fledgling-dependency period. Northern Goshawks 
departed from PFAs 45.9 ± 1.3 d post-fledging. Mean PFA size was 59.2 ± 16.1 ha, and the bootstrapped 
variance around PFA estimates ranged from 12.7-1820.8 ha. Our estimate for the mean size of one PFA 
per nest area for A. g. laingi fledglings on Vancouver Island was much smaller than the mean size 
estimate reported for A. g. atricapillus in New Mexico. However, management plans should consider nest 
areas and PFAs to be one functional component of Northern Goshawk breeding habitat and should 
include multiple alternative nest trees, each with an associated PFA. 

Keywords: Northern Goshawk, Accipiter gentilis laingi; activity centers', adaptive kernel', bootstrapping, fledg- 

ing-dependency period', immature movements', natal dispersal. 


Areas post-emplumamiento de accipiter gentilis laingi uis Vancouver island, Brit- 
ish COLUMBIA 

Resumen. — El area utilizada por las aves inmaduras desde que abandonan el nido hasta que alcanzan 
la independencia de sus padres es el area post-emplumamiento (APE). El unico estimado publicado 
del tamano del APE de Accipiter gentilis (170 ha) corresponde a una poblacion del estado de New Mexico 
(subespecie atricapillus), y la aplicabilidad de este estimado a otras regiones y tipos de habitats es des- 
conocida. Nuestros objetivos fueron estimar el tamano del APE y la longitud del periodo post-emplu- 
mamiento para nidos de A. g. laingi ubicados en Vancouver Island, British Columbia. Estimamos el 
tamano del APE a partir de estimados adaptativos de los kernels del 95% de ubicaciones obtenidas 
mediante telemetria para 12 volantones de 12 nidos, entre el 29 de junio y el 2 de septiembre de 2001- 
02 {N = 6, 2001; N = 6, 2002). Debido a que nuestros estimados de los kernels estan basados en un 
numero pequeno de ubicaciones, tambien estimamos la precision de los estimados del rango de hogar 
empleando un metodo de bootstrap alisado. Casi todas las ubicaciones de los volantones (93%) estu- 
vieron a menos de 200 m de los nidos durante el periodo temprano de emplumamiento-dependencia, 
pero menos de la mitad de las ubicaciones (42.4%) tuvieron lugar a menos de 200 m durante la fase 
tardia de este periodo. Los individuos abandonaron sus APE 45.9 ± 1.3 d despues de abandonar el 
nido. El tamano promedio del APE fue 59.2 ±16.1 ha, y la varianza de los estimados obtenida mediante 


^ Corresponding author’s email address: erica.mcclaren@gov.bc.ca 


253 


254 


McClaren et al. 


VoL. 39, No. 3 


el metodo de bootstrap estuvo entre 12.7 y 1820.8 ha. Nuestro estimado del tamano medio de un APE 
por area de nidificacion para los volantones de A. g. laingi en Vancouver Island es mucho menor que 
el estimado medio documentado para A. g. atricapillus en New Mexico. Sin embargo, los planes de 
manejo deben considerar que las areas de nidificacion y las APE son un componente importante del 
habitat de cria de A. gentilis, y deben incluir multiples arboles que puedan servir como sitios alternativos 
para nidificar, cada uno con su APE asociada. 

[Traduccion del equipo editorial] 


Suitable breeding habitat for avian species con- 
sists of adequate nest sites, roost sites, post-fledging 
areas (PFAs), and foraging areas. PFAs represent 
the habitat used by fledglings prior to indepen- 
dence and may be especially important for species 
with long post-fledging-parental-care periods, such 
as raptors. The survival of fledglings through the 
post-fledging period and their first year is likely in- 
fluenced by PFA quality, which may be reflected by 
PFA size and habitat characteristics. 

Several studies have reported areas around nests 
to be important for fledglings during the post- 
fledging period, before dispersal is initiated (Bald 
Eagles [Haliaeetus kucocephalus] , Wood et al. 1998; 
Great Tits [Parus major], Naef-Daenzer et al. 2001; 
Scarlet Macaws [Ara macao], Myers and Vaughan 
2004). However, the PFA concept (originally re- 
ferred to as the post-fledging family area; Reynolds 
et al. 1992) and its integration with management 
prescriptions (Reynolds et al. 1992) have only been 
applied to Northern Goshawks {Accipiter gentilis) , a 
species of concern in North America (Kennedy 
1997, Crocker-Bedford 1998, DeStefano 1998) and 
Europe (Widen 1997). Kennedy et al. (1994) esti- 
mated the size of goshawk PFAs in New Mexico to 
be ca. 168 ha based on movement patterns of ra- 
dio-tagged fledglings and adult female core-use ar- 
eas. Currently, the British Columbia (BC) govern- 
ment recommends managing a 200-ha PFA around 
designated goshawk {A. g. laingi) nest areas (BC 
Ministry of Water, Land, and Air Protection, 2004). 
This recommendation was a modification of the 
southwestern U.S. guidelines (Reynolds et al. 
1992) because local data were unavailable. 

Our objective was to evaluate the applicability of 
PFA guidelines developed for goshawk populations 
in the southwestern U.S. to coastal BC, where hab- 
itat characteristics, harvest regimes, and goshawk 
subspecies differ. We provide the first PFA estimate 
for goshawks based on home-range estimates of 
fledglings prior to independence as well as preci- 
sion estimates of PFAs, which are rarely provided 
in home-range analyses (Kernohan et al. 2001). 
Local knowledge of goshawk PFA size on Vancou- 


ver Island is crucial for adequately managing the 
breeding habitat of A. g. laingi which is federally 
designated by the Committee on the Status of En- 
dangered Wildlife in Canada (COSEWIC) as 
Threatened in Canada (COSEWIC, in press) and 
is provincially Red-listed (BC Conservation Data 
Centre 2005). 

Methods 

Study Area. Goshawk nest areas were located on Van- 
couver Island, BC (Fig. 1) in forests dominated by west- 
ern hemlock (Tsuga heterophylla) and Douglas-fir (Pseu- 
dotsuga menziesii), although western red cedar {Thuja 
plicata) , amabalis fir {Abies amabalis) , and red alder {Alnus 
rubra) were also abundant. Nest stands ranged in age 
from 45 to >250 yr. See McClaren et al. (2002, 2003) for 
more study area details. 

Data Collection. Goshawk nests {N = 17) used in this 
study were located either through broadcast surveys of 
conspecific calls (McClaren et al. 2002, 2003), or inci- 
dentally by forest company personnel and the public, 
from 1994-2002. When nestlings were ca. 21 d, we 
climbed nest trees and lowered nestlings to the ground 
where they were weighed, measured, sexed, and aged by 
the senior author to maintain consistency in the data. 
Nestling gender was determined using tarsal width, rec- 
ommendations provided by Kenward et al. (1993a). We 
aged nestlings from a photographic and behavioral key 
(Boal 1994) and from our estimated hatch dates. Nest- 
lings were banded with U.S. Geological Survey bands and 
color-rivet bands (Acraft Sign and Nameplate Co. Ltd., 
Edmonton, AB Canada). With two exceptions, we fitted 
only the largest female nestling in each nest with a 9-g 
tarsal mount transmitter with a mortality switch (Ad- 
vanced Technology Services, Isanti, MN U.S.A.). Males 
were fitted with transmitters when: (1) all nestlings were 
male {N = 1); and (2) the sole female nestling’s trans- 
mitter battery died, and we fitted its male sibling with a 
new radiotransmitter. We chose the largest female to re- 
duce potential variation in fledgling movements caused 
by gender differences (Byholm et al. 2003, J. Wiens un- 
publ. data) and to lessen possible impact of transmitter 
mass on survival probability. Radiotransmitters were at- 
tached to tarsi with a leather Jesse (Ward and Kennedy 
1996, Dewey and Kennedy 2001), so that fledglings could 
remove them after the 90 d transmitter battery expired. 
Trade name products are mentioned to provide com- 
plete descriptions of methods. The authors’ institutions 
neither endorse these products nor intend to discrimi- 
nate against products not mentioned. 

Prior to collecting fledgling location data, we centered 
a 600 X 600-m grid on nests with young with location 


September 2005 


Biology 


255 



Map Insert 


AK. USA 


I 


BC. CANADA 


WA. USA 



1:2,100,000 


25 0 25 50 75 



km 


Figure 1. Northern Goshawk nest areas on Vancouver Island, British Columbia, where chicks were captured and 
radiotagged. Small open circles represent nests where there were insufficient radio locations (N < 15) to estimate 
post-fledgling area (PEA) size and small closed circles represent nests where there were sufficient radio locations to 
estimate PFA size. Larger circles around nests represent the 20 km ground search area for fledglings after they began 
to leave the natal area. 



Figure 2. Reference station grid (600 m X 600 m) at 
Northern Goshawk nests on Vancouver Island, British Co- 
lumbia, used for geo-referencing radio-tagged fledgling 
locations from ground-based telemetry during the 2001- 
02 breeding seasons. Each grid cell represents 100 m X 
100 m. 


reference stations at 100-m intervals (Fig. 2). After nest- 
lings were radio-tagged, nests were visited weekly to mon- 
itor chick development, survivorship, and transmitter op- 
eration, Once tagged nestlings fledged, nest areas were 
visited every 1—3 d to collect fledgling location data. We 
rotated sampling equally among nest areas and sampling 
periods (<0800-1100 H, 1101-1400 H, 1401-1700 H, 
and >1700 H) to ensure all times of day were equally 
represented. Teams of two observers listened for radio 
signals prior to entering nest stands. Using a 3-element 
Yagi antenna (Telonics Inc., Mesa, AZ U.S.A.) and re- 
ceiver (Models SRX-1000, SRX-400, Lotek Engineering 
Inc., ON, Canada), observers quietly approached fledg- 
lings on foot to obtain visual locations and to prevent 
influencing their movements. We verified the observed 
fledgling was radio-tagged by identifying color bands. At 
one nest site where the radio-tagged fledgling was pre- 
dated 5 d prior to dispersal, we continued tracking the 
remaining two siblings through aural locations. We mea- 
sured distance and direction of the fledgling to the clos- 
est reference grid station using a meter-marked rope and 
compass. When fledglings became more mobile and 



256 


McClaren et al. 


VoL. 39, No. 3 


moved outside the 600 X 600-m grid, we either expanded 
the grid or used triangulation from roads to estimate 
fledgling locations. We estimated date of departure from 
natal areas as mid-way between the last visit when fledg- 
lings were verified <1.6 km from nests and when no ra- 
dio signal was heard within this distance on two consec- 
utive telemetry sessions (Kenward et al. 1993a, Kennedy 
and Ward 2003). Immediately after young departed from 
natal areas, we conducted intensive road searches for ra- 
dio-tagged fledglings using a vehicle-mounted omni-di- 
rectional antenna (Telonics Inc., Mesa, AZ U.S.A.), and 
when a signal was detected, we used triangulation to ob- 
tain a location. Road searches were conducted within a 
20-km^ area surrounding each nest (Fig. 1). We also 
searched a 30-km area around nests using a single aerial 
telemetry flight. 

Location Determination. We used the survey mapping 
extension of Road Engineering Software (Softree Tech- 
nical Systems Inc. 1998) to calculate UTM coordinates 
for visual fledgling locations based on the known coor- 
dinates of the nest tree location, grid reference station 
locations, and measured offsets from grid reference sta- 
tions to fledglings. Fledgling locations derived from tri- 
angulation and their associated error ellipses were esti- 
mated using Locate II, version 1.5 (Nams 2000), based 
on the number of bearings, angles between bearings, and 
the distance from bearing locations to radio-tagged birds. 

Post-fledging Area Estimation. We used Home Ranger 
(version 1.5, Ursus Software, Revelstoke, BC Canada) to 
estimate PFA size and a smoothed bootstrap (Worton 
1995) with 1000 replications to estimate variance of each 
PFA. Bootstrapping is a common technique for numeri- 
cally estimating the precision of measurements for which 
sampling distributions and variances are unknown (Efron 
and Tibshirani 1991, Quinn and Keough 2002). Al- 
though PFAs are not equivalent to home ranges, we as- 
sumed the area used by fledglings prior to dispersal 
could be modeled using home-range estimation tech- 
niques. Ninety-five percent adaptive kernel estimates 
were used to estimate PFA size because kernel estimators 
were highest ranked in a recent review of the perfor- 
mance of home range estimators (Kernohan et al. 2001). 
Adaptive kernel estimates are non-parametric, indicate 
areas of concentrated use by fledglings (i.e., activity cen- 
ters or core areas), and are more conservative than min- 
imum convex polygon estimators because the home 
range boundaries are based on probability functions 
around bird locations rather than on linear connections 
between the outermost data points (Seaman et al. 1999, 
Kenward et al. 2001). We only included 95% of locations 
because we wanted to exclude exploratory or excursion 
behaviors that could artificially inflate PFA size (Walls 
and Kenward 1998, Kennedy and Ward 2003). We used 
the reference bandwidth ( ^^f) smoothing parameter be- 
cause it appeared to model most accurately the number 
of activity centers in our data. Triangulated locations with 
associated ellipsoid error polygons >1 km^ were not in- 
cluded in PFA estimates {N = 2). We did not estimate 
PFA size for fledglings with <15 locations (N = 3). A 
minimum of 15 locations per fledgling appears small rel- 
ative to the recommended minimum of 30 locations per 
individual for home-range estimation (Kernohan et al. 
2001). However, the short post-fledging period limited 


our ability to collect >30 locations per fledgling that were 
not temporally correlated. In addition, area-observation 
curves (Gese et al. 1990 in Kernohan et al. 2001) sug- 
gested our sample sizes adequately represented maxi- 
mum distances moved from nests during the post-fledg- 
ing period, prior to departure from natal areas. 

Statistical Analyses. We considered fledgling location 
data collected from different nest sites within the same 
nest area in different years to be independent (N = 2). 
Mean and median hatching, fledging, and departure 
from natal area dates were estimated with Julian days. We 
used the correlation procedure in SAS (SAS Institute, 
Inc. 1997) to examine the relationship between fledging 
date and the number of days until dispersal was initiated. 
We used a Fisher’s exact test (PROG FREQ SAS Institute, 
Inc. 1997) to evaluate changes in the distance fledglings 
were from nests at three stages of maturity during the 
post-fledging period. Distances were calculated as the Eu- 
clidian distance from the nest tree to the fledgling loca- 
tion. We categorized the distance we observed fledglings 
(0-99 m, 100-199 m, 200-399 m, 400-799 m, >800 m) 
into three time intervals (1-3 wk, 4—5 wk, and 6-7 wk 
post-fledging). Distance categories were as fine scale as 
possible given the number of locations in each category 
required to run the analyses. We used Wilcoxon’s rank- 
sum tests (PROG NPARIWAY; SAS Institute, Inc. 1997) 
to compare hatching, fledging, dispersal dates, and PFA 
size between years. We used Wilcoxon’s rank-sum tests for 
these pair wise comparisons because small sample sizes 
and non-normal data may invalidate the results provided 
by Kests (Ott 1993). Results were considered statistically 
significant at P < 0.10. Means and standard errors are 
presented unless otherwise stated. Data from all radio- 
tagged fledglings were included in all analyses except 
PFA size estimates and comparisons in which we only in- 
cluded data from young with sufficient locations {N = 
12 ). 

Results 

Forty-two goshawk nestlings from 17 nests were 
banded and measured. Mean age of young at 
banding was 20.2 ± 0.8 d. Tarsus width for male 
(N = 15) and female {N = 27) chicks averaged 6.1 
±0.1 mm and 7.3 ± 0.1 mm, respectively, and 
means were significantly different (^s “ 2.0, P < 
0 . 001 ). 

Breeding Phenology. Median hatch date was 29 
May (N = 17), median fledge date was 7 July (N 
= 17), and median date of departure from PFAs 
was 25 August (N = 15). Goshawk nestlings spent 
a mean of 40.4 ± 0.3 d (N = 17) in their nests 
before fledging and 45.9 ± 1.3 d {N = 15) in PFAs 
before departing from natal areas. Goshawks 
hatched (S = 65.0, P = 0.07) and left PFAs (5 = 
24.5, P = 0.003) significantly earlier in 2002 than 
in 2001. In 2001, individuals that fledged later 
spent less time within PFAs than early fledged 
young (Fig. 3). Thus, hatching date (r = —0.8, P 


September 2005 


Biology 


257 



60 1 

< 

u. 

55 - 

Q. 

E 

50 

o 

w 

Li. 

P 

45 - 

tr 

03 

40 ■ 

Q. 

0) 

Q 

35 - 



170 


180 190 200 

Julian Day of Fledging 


Figure 3. Relationship between fledging date and time 
spent within post-fledgling areas (PFAs) before dispersal 
initiation for Northern Goshawk fledglings on Vancouver 
Island, British Columbia, in 2001 and 2002. 


= 0.007) and fledging date (r = -0.8, P = 0.01; 
Fig. 3) were negatively correlated with the amount 
of time young spent in PFAs before initiating dis- 
persal. In 2002, relationships between hatching 
date (r = -0.3, P = 0.5) and fledging date (r = 
-0.3, P = 0.5; Fig. 3) with the amount of time 
young spent within PFAs were weak because both 
early and late fledged young spent nearly equal 
time periods within PFAs. In both years, fledgling 
departure from PFAs occurred over a 2-wk period 
(2001: 10-25 August; 2002: 20 August-2 Septem- 
ber) . 

Fledgling Location Data. We collected 236 radio- 
telemetry locations from 15 radio-tagged goshawk 
fledglings. Most of these were visual locations 
(93.2% N = 220). Triangulated locations (6.8%; N 
= 16) had a mean error ellipse of 0.029 ± 0.009 
km^, equivalent to a circle around each location 
with a mean radius of 96.1 m. Ninety-three percent 
of fledgling locations were within 200 m of nests 
{x = 107.8 ± 8.9 m, iV = 105) during the first 3 
wk post-fledging, but only 42.4% of locations were 
within this distance (x = 261 ± 17.5 m, N = 131) 
during the remaining 4 wk post-fledging (Fig. 4) . 
As fledglings matured, we generally located them 
farther from nests (2001: 52.5, P < 0.001; 

2002: - 32.4, P< 0.001). However, we observed 

fledglings returning to nest trees throughout the 
post-fledging period, so fledglings did not contin- 
ue to expand their PFA size indefinitely until they 
departed from natal areas. Maximum movements 
were observed in the 2 wk after fledglings com- 
pleted feather growth and subsequent feather 
hardening, ca. 70-75 d of age (Fig. 4). We were 


1400 


1200 - Fledging 


Feather 

Hardening 


00% Dispersal 
Complete 


-P 1000 


800 
600 - 
400 - 
200 


35 



95 


Figure 4. Distance of Northern Goshawk fledglings {N 
= 15) from nests during the post-fledging period on Van- 
couver Island, British Columbia, 29 June-2 September, 
2001-02. Vertical lines represent median fledge date and 
estimated median feather-hardening date. Horizontal 
grey bars represent mean distances fledglings were ob- 
served from nests pre-feather hardening and post-feather 
hardening. 


only able to relocate one fledgling 27 d after it left 
the PFA, and it was 82 km from its natal site. 

We gathered 17.8 ± 0.6 (range = 15—22) loca- 
tions per fledgling from the 12 fledglings for which 
we had sufficient locations {N > 15) to estimate 
PFA size (Table 1). We did not include locations 
from one male fledgling because he was sick and 
remained near the same perch tree until he was 
recovered dead 3 wk post-fledging. Overall, mean 
PFA size was 59.2 ± 16.1 ha {N = 12), and the 
range was 14.5-229.7 ha (Table 1). Variance 
around PFA estimates based on the bootstrap sam- 
ples ranged from 12.7—1820.7 ha (Table 1). PFA 
size did not significantly differ {S = 39.0, P = 1.0) 
between 2001 (71.1 ± 32.3 ha) and 2002 (47.4 ± 
6.8 ha), although small sample sizes may have re- 
duced our power to detect annual differences. 

The size and shape of PFAs varied among fledg- 
lings, and one fledgling used >1 activity centers. 
In 75% of PFAs, fledgling activity centers included 
nest trees. The three fledglings with PFAs that did 
not include nest trees had activity centers that were 
ca. 100 m, 150 m, and 300 m from nest trees. 

Fledgling Fate. Brood reduction occurred at a 
minimum of seven nests prior to banding. At least 
three nestlings died post-banding, but pre-fledg- 
ing. In 2001, 100% of radio-tagged fledglings sur- 
vived through the post-fledging period (Table 2). 


258 


McClaren et al. 


VoL. 39, No. 3 


Table 1. Post-fledging area estimates for Northern Goshawk nests (N = 12) on Vancouver Island, British Columbia, 
29 June-2 September 2001-02. 


Nest Area 



Bootstrap 
Min., Max. 

Percent Visual 
Locations 

N 

Name 

War 

PFA Size (ha) 

Great Central Lake 

2001 

54.3 

50.8, 173.8 

69 

17 

Klaklakama No. 7 

2001 

229.7 

182.7, 1820.8 

53 

20 

Loon Lake 

2001 

34.7 

26.4, 275.9 

100 

18 

Paterson 

2001 

53.4 

39.0, 446.3 

95 

21 

Roberts Lake 

2001 

14.5 

12.7, 101.0 

95 

23 

Toad No. 2 

2001 

40.0 

36.4, 107.6 

100 

20 

Sutton 

2002 

26.7 

24.5, 103.0 

100 

15 

John Road 

2002 

77.8 

36.2, 1120.0 

100 

17 

Klaklakama No. 3 

2002 

46.6 

32.0, 272.5 

100 

19 

Toad No. 3 

2002 

47.9 

23.4, 547.8 

100 

16 

Surprise Lake 

2002 

40.9 

34.0, 259.5 

100 

16 

Pye Lake 

2002 

44.4 

23.7, 372.1 

100 

17 


However, in 2002, 37.5% {N = 8) of radio-tagged 
fledglings, and at least one untagged fledgling, 
died prior to leaving the natal area. We experi- 
enced one premature battery failure, and one 
fledgling removed its transmitter prior to initiating 
dispersal. 

Discussion 

Post-fledging Period Behavior. Goshawk fledg- 
lings on Vancouver Island exhibited similar move- 
ment patterns during the post-fledging period to 
goshawks in Sweden (Kenward et al. 1993a) and 
New Mexico (Kennedy et al. 1994, Kennedy and 
Ward 2003). Within the first 3 wk post-fledging, 
fledglings remained within 200-300 m (x = 107.8 
± 8.9 m, N = 105) of nests, and they were often 
on the ground or in the lower canopy. Immediately 
after this period, fledglings experienced a behav- 
ioral transition in which they were frequently lo- 
cated farther from nests (x = 261 ± 17.5 m, V = 
131); they were adept fliers and perched in the 
upper canopy, often in treetops. These changes in 
fledgling behavior correspond with completion of 
primary and retrix feather growth and subsequent 
feather hardening (Kenward et al. 1993a). Inter- 
estingly, goshawks did not continue to expand 
their PFA size indefinitely until departing from 
nest areas. Instead, the farthest distance we ob- 
served fledglings from nests during the post-fledg- 
ing period peaked within 1-2 wk after they com- 
pleted feather growth and approximately 10 d 
prior to departing PFAs. Kenward et al. (1993a) 
and Minguez et al. (2001) described a similar pat- 


tern for goshawks in Sweden and for Bonelli’s 
Eagles (Hieraaetus fasciatus) in Spain, respectively. 
Such a pattern illustrates the importance of col- 
lecting fledgling locations uniformly throughout 
the post-fledging period when trying to character- 
ize fledgling movement patterns. 

In 2001, hatch and fledge dates were negatively 
correlated vdth the amount of time young spent 
within PFAs. Similar negative relationships between 
hatch date and age when dispersal was initiated 
were reported for goshawks in Sweden (Kenward 
et al. 1993b) and Finland (Byholm et al. 2003). 
However, in 2002 goshawks initiated breeding ear- 
lier on Vancouver Island than in 2001, and fledg- 
lings spent similar amounts of time within PFAs, 
regardless of their hatch and fledge dates. In sev- 
eral bird species, an early onset of breeding often 
indicates higher food availability within the nest 
area, which results in higher fledgling mass and 
survival (Dewey and Kennedy 2001, Naef-Daenzer 
et al. 2001, Aparicio and Bonal 2002). Because we 
did not manipulate any proximate factors that may 
have influenced the length of the post-fledging pe- 
riod, we can only speculate on what influenced the 
timing of fledgling departure from natal areas and 
length of post-fledging periods in our study. Food 
availability within home ranges, predator and com- 
petitor abundance, and weather are all possible in- 
fluential factors (Kenward et al. 1993a, Dewey and 
Kennedy 2001, Byholm et al, 2003). We found no 
evidence that parental aggression caused fledglings 
to disperse on Vancouver Island, which supports 


September 2005 


Biology 


259 


Table 2. Fate of radio-tagged Northern Goshawk fledglings 29 June-11 September 2001-02 during post-fledgling 
area (PFA) estimation on Vancouver Island, British Columbia. 


Fledgling 

ID 

Year Tagged 

Date Last Obs. 
(Bird Age in d) 

Fate 

China 

2001 

27-Aug-Ol (84) 

Departed from PFA 

Cous 

2001 

22-Aug-Ol (88) 

Departed from PFA 

Great Central Lake 

2001 

27-Aug-Ol (79) 

Departed from PFA 

Mesachie 

2001 

16-Aug-Ol (89) 

Departed from PFA 

Klaklakama No. 7 

2001 

24-Aug-Ol (94) 

Departed from PFA 

Loon Lake No. 3 

2001 

24-Aug-Ol (88) 

Departed from PFA 

Paterson 

2001 

25-Aug-Ol (82) 

Departed from PFA 

Roberts Lake 

2001 

25-Aug-Ol (97) 

Departed from PFA 

Toad No. 2 

2001 

24-Aug-Ol (90) 

Departed from PFA 

Claud Elliot 

2002 

29-Jul-02 (63) 

Dead (unknown cause) 

Toad No. 3 

2002 

05-Aug-02 (77) 

Dead (predated) 

John Road a^ 

2002 

03-Aug-02 (75) 

Battery failed 

John Road b 

2002 

ll-Sep-02 (114) 

Mortality switch on 

Loon Lake No. 3 

2002 

14-July-02 (41) 

Dead (unknown cause) 

Sutton 

2002 

lO-Aug-02 (79) 

Departed from PFA 

Klaklakama No. 3 

2002 

19-Aug-02 (87) 

Departed from PFA 

Pye Lake 

2002 

21-Aug-02 (87) 

Departed from PFA 

Surprise Lake 

2002 

23-Aug-02 (83) 

Departed from PFA 


® Two individuals were radiotagged at this nest because the first radiotransmitter battery died. We captured and radiotagged its sibling 
after a failed attempt to recapture the originally tagged individual. 


experimental results provided by Kenward et al. 
(1993a). 

Although the onset of fledgling dispersal varied 
by approximately 10 d between 2001 and 2002, 
fledglings departed natal areas abruptly between 
80-96 d of age in both years. This seems to be a 
common pattern for goshawks (Kenward et al. 
1993a, Dewey and Kennedy 2001) and for many 
other raptors (Spotted Owls [Strix occidentalism. Wil- 
ley and van Riper 2000; Bonelli’s Eagles: Minguez 
et al. 2001). In contrast, Walls and Kenward (1998) 
reported a bimodal pattern of departure from na- 
tal areas for Common Buzzards {Buteo buteo) and 
Kennedy and Ward (2003) observed supplemen- 
tally-fed goshawk fledglings returning to natal ar- 
eas after they initiated dispersal. We were unable 
to evaluate movement patterns for radio-tagged 
goshawks during their first year of life because our 
transmitter batteries expired when young were ca. 
110 d of age. However, we searched for fledglings 
within 30 km of nest areas after they initiated dis- 
persal, and our inability to locate them suggested 
fledglings moved >30 km after departing PFAs. 
Initial departure distances were probably moder- 
ated by local food availability, whereby fledglings 
within food-rich areas moved shorter distances af- 


ter leaving natal areas than fledglings from food- 
poor areas (Kenward et al. 1993b, Kennedy and 
Ward 2003). 

Post-fledging Area Size. Most fledglings included 
nest trees within their activity centers throughout 
the post-fledging period. Similar patterns reported 
by Ward and Kennedy (1996: goshawks). Wood et 
al. (1998: Bald Eagles), and Belthoff and Ritchison 
(1989: Eastern Screech-Owls [Otus asio^) suggest 
that nest trees are important throughout the post- 
fledging period for raptors. Some goshawk man- 
agement guidelines recommend reduced distur- 
bance levels around goshawk nests until young 
fledge (e.g., BC Ministry of Water, Land, and Air 
Protection 2004). However, the vulnerability of 
young during the early fledging-dependency peri- 
od (Wiens 2004) and their continued use of the 
nest site throughout the post-fledging period, sug- 
gests there should be strict adherence to distur- 
bance recommendations until young leave PFAs. 
Disturbance near nest areas during the post-fledg- 
ing period may interfere with adult prey deliveries 
to young and development of juvenile hunting and 
flight skills (Kenward et al. 1993a, Kennedy et al. 
1994, Wood et al. 1998). 

Our study reports the first published estimate of 


260 


McClaren et al. 


VoL. 39, No. 3 


goshawk PFAs based on home-range estimates de- 
rived almost entirely from visual locations, with no 
location error. Our estimated mean PFA size of 
59.2 ha is smaller than that reported by Kennedy 
et al. (1994). Kennedy et al. (1994) based their PFA 
size estimate on adult female core use areas which 
were corroborated with fledgling location data, 
rather than calculating PFA size directly from 
fledgling locations. Also, reanalysis of fledgling lo- 
cation data from Kennedy et al.’s (1994) study in- 
dicated the non-visual observations of fledglings 
frequently had a 500-m radius error (Kennedy and 
Ward 2003). Similarly, fledgling distances provided 
by Kenward et al. (1993a) were likely inflated be- 
cause their telemetry locations were accurate with- 
in only 100 m. Our PFA estimates may have been 
slightly inflated because they included one fledg- 
ling (Klaklakama No. 7) from which 47% of loca- 
tions were collected using triangulation with an as- 
sociated 104.4-m radius error (Table 1). However, 
our second largest PFA estimate was for a fledgling 
(John Road) for which 100% of locations were vi- 
suals. 

Additionally, post-fledging movement patterns 
may be influenced by fledgling gender (Byholm et 
al. 2003, J. Wiens unpubl. data) and by landscape 
habitat characteristics surrounding nests. Our PFA 
estimates may be smaller than those reported by 
Kennedy et al. (1994) because all but one PFA es- 
timate were for females that, in one northern Ar- 
izona study, were smaller than male PFAs (J. Wiens 
unpubl. data). PFA size on Vancouver Island may 
also be smaller than in New Mexico because the 
definitive forest edges of nest stands in coastal for- 
est ecosystems may act as barriers to fledgling 
movements more than the less defined ecotones 
that occur between southwestern forest types (Sid- 
ers and Kennedy 1996). 

Because PFA size can only be estimated from lo- 
cation data, providing variance estimates for these 
and other types of home range data is extremely 
informative, but rarely done (Worton 1995, Ker- 
nohan et al. 2001). Our variance estimates of the 
PFA estimates include the 169 ± 129 ha PFA size 
reported by Kennedy et al. (1994) and the dis- 
tances (100-1000 m) that Kenward et al. (1993a) 
observed fledglings from nests. Our PFA estimate 
was closer to the minimum bootstrapped estimate 
than the maximum bootstrapped estimate because 
a greater proportion of our location data were clos- 
er to nests. Few fledgling locations far from nests 
created more variability in the maximum boot- 


strapped estimate, although maximum estimates 
provide important information (F. Hovey pers. 
comm.). 

Home-range estimates also vary depending on 
the techniques used to collect location data and 
on the home-range estimation program used to 
calculate home range size (Lawson and Rodgers 
1997, Seaman et al. 1999, Kenward et al. 2001). For 
example, Kennedy et al. (1994) used a harmonic 
mean estimator to calculate female core use areas, 
whereas we used an adaptive kernel estimate. 
Therefore, comparison of PFA size estimates 
among studies that use different data collection 
and size estimation techniques is difficult. 

Management Implications. Most goshawk man- 
agement guidelines in North America are based on 
Reynolds et al. (1992), which suggest managing for 
three hierarchical levels of goshawk home ranges: 
(1) nest area, (2) PFA, and (3) foraging area. How- 
ever, Reynolds et al. (1992) also recommended 
managing for alternative nests within goshawk nest 
areas and they assumed that all alternative nests 
were within PFAs. Therefore, the biological func- 
tionality of a nest area independent of a PFA is 
questionable, and managing these habitat compo- 
nents in isolation may reduce the effectiveness of 
management plans. Recent studies comparing hab- 
itat characteristics around goshawk nests to ran- 
dom sites (areas assumed not to contain goshawk 
nests) at multiple spatial scales concluded that gos- 
hawk habitat could be discriminated from random 
sites by a larger proportion of large-diameter, late- 
seral, closed-canopy forests (Ethier 1999) at scales 
between 83 ha (McGrath et al. 2003) and 170 ha 
(Daw and DeStefano 2001). Additionally, Finn et 
al. (2002) reported occupied historic goshawk 
nests had a greater proportion of late-seral forest 
with high canopy closure, less stand initiation cov- 
er, and reduced landscape heterogeneity at 177 ha 
and 1886 ha scales, than at similar scales around 
unoccupied historic nests. These studies suggest 
goshawk PFAs may be characterized by unique hab- 
itat characteristics at spatial scales within the size 
range we have reported for PFAs as well as the size 
range reported by Kennedy et al. (1994). 

Mean PFA size estimates on Vancouver Island 
were smaller than the 200-ha area currently rec- 
ommended for managing the area around gos- 
hawk nests in coastal BC (BC Ministry of Water, 
Land, and Air Protection 2004). However, our re- 
sults represent only one nest, and therefore, one 
PFA per nest area, within a given year. We moni- 


September 2005 


Biology 


261 


tored fledglings from two different nest sites within 
two nest areas in 2001 and 2002, and there was 
minimal overlap between PFAs in different years. 
This suggests that each alternative nest site may 
have a unique PFA. Therefore, a more meaningful 
approach to managing goshawk breeding home 
ranges is to manage for areas that include multiple 
nests and associated PFAs. Our bootstrapping re- 
sults suggest this PFA size is highly variable and 
likely depends upon methods used to estimate PFA 
size as well as environmental factors such as topog- 
raphy, habitat characteristics around nests, prey 
availability, and fledgling gender (Dewey and Ken- 
nedy 2001, Byholm et al. 2003, Kennedy and Ward 
2003, J. Wiens, unpubl. data). 

We developed a simplistic graphical depiction of 
how our information could be applied to develop 
management scenarios for A. g. laingi nest areas 
throughout coastal BC (Fig. 5) . This figure is based 
on Vancouver Island data with a mean number of 
3.0 ± 0.2 (N = 34 nest areas) alternative nests/ 
nest area and a mean distance of 274 ± 37.2 m (N 
— 65) between alternative nest trees (E. McClaren 
unpubl. data) . The total area to be managed would 
vary by nest area and depends on the juxtaposition 
of alternative nests and PFAs (Fig. 5). For example, 
Figures 5a and 5b depict areas that are 104.8 ha 
and 96.3 ha in size, respectively. In areas where the 
inter-alternate distance is larger (inter-alternate 
distances >1.0 km are not uncommon; Dewey et 
al. 2003, Squires and Kennedy in press), the total 
management area would be larger. In the absence 
of fledgling radiotelemetry data and information 
on fledgling habitat selection patterns, multiple 
PFAs within one goshawk home range should be 
managed to create an area that maintains connec- 
tivity among alternative nests and to adjacent 
stands of similar habitat (i.e., reduce stand isola- 
tion) to minimize possible edge effects, facilitate 
food transfers from adults, and provide dispersal 
corridors. 

Although our results suggested the area used by 
goshawk fledglings on Vancouver Island, and pos- 
sibly elsewhere, was smaller than estimated in New 
Mexico (Kennedy et al. 1994), PFA habitat was not 
the only habitat necessary for goshawks to success- 
fully reproduce. Prey availability in habitats outside 
of, but in proximity to, PFAs was also essential for 
adults to rear young (Reynolds et al. 1992, Ken- 
nedy and Ward 2003, Wiens 2004). For example, 
Bloxton (2002) reported radio-tagged adult gos- 
hawks in the Olympic Peninsula, WA, to concen- 



Figure 5. Conceptual representation of managing three 
alternative Northern Goshawk nests and their associated 
post- fledgling areas (PFAs) . Two possible configurations, 
(a) all nests adjacent or (b) two adjacent and one below, 
are shown with 274 m as the mean distance between al- 
ternative nests and a mean PFA size of 59 ha on Vancou- 
ver Island, British Columbia. The diagram is drawn to 
scale. 


trate their foraging efforts within 5 km of occupied 
nests during the breeding season. Current goshawk 
management guidelines in BC (BC Ministry of Wa- 
ter, Land and Air Protection 2004) do not include 
explicitly managing for goshawk foraging areas, 
and the effect of not managing goshawk foraging 
areas in landscapes actively managed for timber 
harvest is unknown. 

Research Recommendations. In future PFA stud- 
ies, we recommend increasing the minimum num- 
ber of locations/ fledgling to a minimum of 30 to 
improve the precision of PFA estimates (Kernohan 
et al. 2001). Goshawk post-fledging periods are ex- 
tremely short, and the timeframe for data collec- 
tion is limited. Obtaining reasonable samples of lo- 
cations will require collecting either daily locations 


262 


McClaren et al. 


VoL. 39, No. 3 


after young fledge or collecting multiple locations 
per sample day and relaxing guidelines around in- 
dependence of locations. A sampling regime that 
spaces location data collection evenly throughout 
the duration of the study, enabling a reasonable 
amount of time for animals to relocate, may be 
more important than concerns about autocorrela- 
tion of data (Kernohan et al. 2001), Additionally, 
PFA size and habitat studies should be conducted 
across a diversity of ecosystems, so that manage- 
ment recommendations may be fine-tuned to re- 
flect similarities and differences across broad geo- 
graphic areas. This information may assist with 
designating suitable PFAs around nests when it is 
not feasible to collect radiotelemetry data. 

Acknowledgments 

We appreciate Forest Renewal BC and the Forest In- 
vestment Account funding allocated by Canadian Forest 
Products, Weyerhaeuser, TimberWest and Western Forest 
Products, as well as the cooperation of these companies, 
enabling us to work on their tree farm licenses and pri- 
vate lands. Pacific Northwest Canadian Studies Consor- 
tium provided assistance with travel for authors to collab- 
orate. This project would not have been possible without 
the labor-intensive data collection provided by; J. Bond, 
N. Davey, R. Dickson, I. Jacobs, M. Krkosek, P. Levesque, 
J. Malt, C. Pendergast, J. Strain, and A. Zeeman. We ex- 
tend our gratitude to I. Jacobs for fearlessly climbing gos- 
hawk nest trees and to C. Pendergast for assisting with 
data organization and the figures and tables in this re- 
port. T. Dunlop assisted with the CIS work and calcula- 
tions for the conceptual PFA diagram. We thank D. Cator 
for keeping her eyes and ears cued for goshawks while 
canoeing. S. Dewey, F. Doyle and J. Ward provided valu- 
able technical advice for this project. Discussions with J. 
Horne, F. Hovey, J. Millspaugh, and A. Rodgers clarified 
our understanding of home-range estimation programs 
and conducting bootstrapping within home-range pro- 
grams. J. Morrison and two anonymous reviewers im- 
proved the quality of this manuscript with their editorial 
suggestions. 

Literature Cited 

Aparicio, J.M. and R. Bonal. 2002. Effects of food sup- 
plementation and habitat selection on timing of Less- 
er Kestrel breeding. Ecology 83:873—877. 

Belthoff, J.R. and G. Ritchison. 1989. Natal dispersal 
of Eastern Screech-Owls. Condor 91:254— 265. 
Bloxton, T.D., Jr. 2002. Prey abundance, space use, de- 
mography, and foraging habitat of Northern Gos- 
hawks in western Washington. M.S. thesis, Univ. Wash- 
ington, Seattle, WA U.S.A. 

Boat. C.W. 1994. A photographic and behavioral guide 
to aging nestling Northern Goshawks. Stud. Avian 
Biol. 16:32-40. 

British Columbia Ministry of Water, Land, and Air 
Protection. 2004. Identified wildlife management 


strategy accounts and measures for managing identi- 
fied wildlife. Available: http://wlapwww.gov.bc.ca/ 
wld/identified/accounts.html (27 June 2005). 

British Columbia Conservation Data Centre. 2005 
British Columbia species and ecosystem explorer. Brit- 
ish Columbia Ministry of Sustainable Resource Man- 
agement. Victoria, BC Canada. Available: <http:// 
srmapps.gov.bc.ca/apps/eswp/ (June 27, 2005). 

Byholm, R, P. Saurola, H. Linden, and M. Wikman. 
2003. Causes of dispersal in Northern Goshawks {Ac- 
dpiter gentilis) banded in Finland. Auk 120:706-716. 

Committee on the Status of Endangered Wildlife in 
Canada. In press. COSEWIC assessment and update 
status report on the Northern Goshawk lain^ subspe- 
cies Accipiter gentilis laingi in Canada. Committee on 
the Status of Endangered Wildlife in Canada, Ottawa, 
ON Canada. 

Crocker-Bedford, D.C. 1998. The value of demographic 
and habitat studies in determining the status of 
Northern Goshawks {Acdpiter gentilis atricapillus) with 
special reference to Crocker-Bedford (1990) and Ken- 
nedy (1997)./. Raptor Res. 32:329-335. 

Daw, S.K. and S. DeStefano. 2001. Forest characteristics 
of Northern Goshawk nest stands and post-fledging 
family areas in Oregon. J. Wildl. Manag. 65:46-58. 

DeStefano, S. 1998. Determining the status of Northern 
Goshawks in the west: is our conceptual model cor- 
rect? J. Raptor Res. 32:342-348. 

Dewey, S.R. and P.L. Kennedy. 2001. Effects of supple- 
mental food on parental-care strategies and juvenile 
survival of Northern Goshawks. Auk 118:352-365. 

-, , and R.M. Stephens. 2003. Are dawn vo- 
calization surveys effective for monitoring goshawk 
territory occupancy? J. Wildl. Manag. 67:390-397. 

Efron, B. and R. Tibshirani. 1991. Statistical data anal- 
ysis in the computer age. Sdence 253:390-395. 

Ethier, T. 1999. Breeding habitat of Northern Goshawks 
{Acdpiter gentilis laingi) on Vancouver Island: a hier- 
archical approach. M.S. thesis, Univ. Victoria, Victoria, 
BC, Canada. 

Finn, S.R, D.E. Varland, and J.M. Ma rzl uff. 2002. Does 
Northern Goshawk breeding occupancy vary with 
nest-stand characteristics on the Olympic Peninsula, 
Washington? /. Raptor Res. 36:265-279. 

Kennedy, P.L. 1997. The Northern Goshawk {Acdpiter gen- 
tilis atricapillus) : is there evidence of a population de- 
cline? J. Raptor Res. 31:95-106. 

and J.M. Ward. 2003. Effects of experimental 

food supplementation on movements of juvenile 
Northern Goshawk {Acdpiter gentilis atricapillus). Oec- 
ologia 134:284-291. 

, , G.A. Rinker, and J.A. Gessaman. 1994. 

Post-fledging areas in Northern Goshawk home rang- 
es. Stud. Avian Biol. 16:75-82. 

Kenward, R.E., R.T. Clarke, K.H. Hodder, and S. Walls. 
2001. Density and linkage estimators of home range. 


September 2005 


Biology 


263 


nearest-neighbor clustering defines multinuclear 
cores. Ecology 82:1905-1920. 

, V. Marcstrom, and M. Karlbom. 1993a. Post- 
nestling behaviour in goshawks, Accipiter gentilis: the 
causes of dispersal. Anim. Behav. 46:365-370. 

, , AND . 1993b. Post-nestling behav- 
iour in goshawks, Accipiter gentilis: II. Sex differences 
in sociality and nest-switching. Anim. Behav. 46:371- 
378. 

Kernohan, B.J., R.A. Gitzen, and J.J. Millspaugh. 2001. 
Analysis of animal space use and movements. Pages 
125-166 wJ.J. Millspaugh and J.M. Marzluff [Eds.], 
Radio tracking and animal populations. Academic 
Press, San Diego, CA U.S.A. 

Lawson, E.J.G. and A.R. Rodgers. 1997. Differences in 
home-range size computed in commonly used soft- 
ware programs. Wildl. Soc. Bull. 25:721-729. 

McClaren, E.L., P.L. Kennedy, and P.L. Chapman. 2003. 
Efficacy of male goshawk food-delivery calls in broad- 
cast surveys on Vancouver Island. J. Raptor Res. 37: 
198-208. 

, , AND S.R. Dewey. 2002. Do some North- 
ern Goshawk nest areas consistently fledge more 
young than others? Condor 104:343—352. 

McGrath, M.T., S. DeStefano, R.A. Riggs, L.L. Irwin, 
AND G.J. Roloff. 2003. Spatially explicit influences on 
Northern Goshawk nesting habitat in the interior Pa- 
cific northwest. Wildl. Monogr. 154:1-63. 

Minguez, E., E. Angulo, and V. Siebering. 2001. Factors 
influencing length of the post-fledging period and 
timing of dispersal in Bonelli’s Eagle {Hieraaetus fas- 
ciatus) in southwestern Spain. 2001. f. Raptor Res. 35: 
228-234. 

Myers, M.C. and C. Vaughan. 2004. Movement and be- 
havior of Scarlet Macaws {Ara macao) during the post- 
fledging dependence period: implications for in situ 
versus ex situ management. Biol. Conserv. 118:411- 
420. 

Naef-Daenzer, B., F. Widmer, and M. Nuber. 2001. Dif- 
ferential post-fledging survival of Great and Coal Tits 
in relation to their condition and fledging date. J. 
Anim. Ecol. 70:730-738. 

Nams, VO. 2000. Locate II user’s guide. Pacer, Truro, 
Nova Scotia, Canada. 

Ott, R.L. 1993. An introduction to statistical methods 
and data analysis, 4th Ed. Duxbury Press, Belmont, 
CA U.S.A. 

Quinn, G.P. and M.J. Keough. 2002. Experimental de- 


sign and data analysis for biologists. Cambridge Univ. 
Press, Cambridge U.K. 

Reynolds, R.T., R.T. Graham, M.H. Reiser, R.L. Bassett, 
P.L. Kennedy, D.A. Boyce, G. Goodwin, R. Smith, and 
E.L. Fisher. 1992. Management recommendations for 
the Northern Goshawk in the southwestern United 
States. USDA Forest Service General Technical Re- 
port RM-GTR-217, Fort Collins, CO U.S.A. 

SAS Institute, Inc. 1997. SAS/STAT Software: changes 
and enhancements through release of 6.12. SAS In- 
stitute, Inc., Cary, NC USA. 

Seaman, D.E., JJ- MILI.SPAUGH, B.J. Kernohan, G.C. 
Brundige, K.J. Raedeke, and R.A. Gitzen. 1999. Ef- 
fects of sample size on kernel home range estimates. 
J. Wildl. Manag. 63:739-747. 

SiDERS, M.S. AND P.L. Kennedy. 1996. Forest structural 
characteristics of Accipiter nesting habitat: is there an 
allometric relationship? Corerfor 98:123-132. 

SoFTREE Technical Systems Inc. 1998. Road Eng forest 
engineer vs. 3.0. Softree Technical Systems, Inc., West 
Vancouver, BC Canada. 

Squires, J. and P.L. Kennedy. In press. Northern Gos- 
hawk ecology: an assessment of current knowledge 
and information needs for conservation and manage- 
ment. Stud. Avian Biol. 

Walls, S.S. and R.E. Kenward. 1998. Movements of ra- 
diotagged buzzards Buteo buteo in early life. Ibis 140' 
561-568. 

Ward, J.M. and P.L. Kennedy. 1996. Effects of supple- 
mental food on size and survival of juvenile Northern 
Goshawks. Auk 113:200-208. 

Widen, P. 1997. How, and why, is the goshawk {Accipiter 
gentilis) affected by modern forest management in 
Fennoscandia? J. Raptor Res. 31:107-113. 

Wiens, J.D. 2004. Post-fledging survival and natal dis- 
persal of Northern Goshawks in Arizona. M.S. thesis, 
Colorado State University, Fort Collins, CO U.S.A. 

Willey, D.W. and C. van Riper, III. 2000. First-year move- 
ments by juvenile Mexican Spotted Owls in the Can- 
yonlands of Utah./. Raptor Res. 34:1-7. 

Wood, P.B., M.W. Collopy, and C.M. Sekerak. 1998. 
Postfledging nest dependence period for Bald Eagles 
in Florida./. Wildl. Manag. 62:333-339. 

WORTON, BJ- 1995. Using Monte Carlo simulation to 
evaluate kernel-based home range estimators./. Wildl. 
Manag. 59:794—800. 

Received 2 February 2004; accepted 2 September 2004 

Guest Editor: Stephen DeStefano 


J. Raptor Res. 39(3):264— 273 
© 2005 The Raptor Research Foundation, Inc. 


NORTHERN GOSHAWK DIET IN MINNESOTA: 

AN ANALYSIS USING VIDEO RECORDING SYSTEMS 

Brett L. Smithers^ 

Department of Range, Wildlife and Fisheries Management, Texas Tech University, 

Lubbock, TX 79409-2120 US. A. 

Clint W. Boat 

us. Geological Survey, Texas Cooperative Fish and Wildlife Research Unit, Texas Tech University, 

Lubbock, TX 79409-2120 U.S.A. 

David E. Andersen 

U.S. Geological Survey, Minnesota Cooperative Fish and Wildlife Research Unit, University of Minnesota, 

St. Paul, MN 55108-6124 U.S.A. 


Abstract. — ^We used video-recording systems to collect diet information at 13 Northern Goshawk (Ac- 
cipiter gentilis) nests in Minnesota during the 2000, 2001, and 2002 breeding seasons. We collected 4871 
hr of video footage, from which 652 prey deliveries were recorded. The m^ority of prey deliveries 
identified were mammals (62%), whereas birds (38%) composed a smaller proportion of diet. Mammals 
accounted for 61% of biomass delivered, and avian prey items accounted for 39% of prey biomass. 
Sciurids and leporids accounted for 70% of the identified prey. Red squirrel ( Tamiasciurus hudsonicus ) , 
eastern chipmunk ( Tamias striatus) , and snowshoe hare (Lepus americanus) were the dominant mammals 
identified in the diet, while American Crow {Corvus brachyrhynchos) and Ruffed Grouse {Bonasa umbellus) 
were the dominant avian prey delivered to nests. On average, breeding goshawks delivered 2.12 prey 
items/d, and each delivery averaged 275 g for a total of 551 g delivered/d. However, daily {P< 0.001) 
and hourly {P — 0.01) delivery rates varied among nests. Delivery rates (P = 0.01) and biomass delivered 
{P = 0.038) increased with brood size. Diversity and equitability of prey used was similar among nests 
and was low throughout the study area, most likely due to the dominance of red squirrel in the diet. 

EIey Words: Northern Goshawk, Accipiter gentilis; diet, Minnesota', prey diversity, red squirrel, Tamiasciurus 

hudsonicus. 


DIETA DE ACCIPITER GENTILIS EN MINNESOTA: UN ANALISIS BASADO EN SISTEMAS DE GRA- 
BACION EN VIDEO 

Resumen. — Empleamos sistemas de grabacion en video para recolectar informacion sobre la dieta de 
Accipiter gentilis en 13 nidos ubicados en Minnesota durante las temporadas reproductivas de 2000, 2001 
y 2002. Obtuvimos 4871 hr de grabacion, a partir de las cuales registramos 652 entregas de presas. La 
mayoria de las presas entregadas que identificamos fueron mamiferos (62%), mientras que las aves 
(38%) representaron una proporcion menor de la dieta. Los mamiferos y las aves representaron el 61% 
y el 39% de la biomasa entregada, respectivamente. Los sciuridos y leporidos representaron el 70% de 
las presas identificadas. Los mamiferos predominantes identificados en la dieta fueron Tamiasciurus 
hudsonicus, Tamias striatus y Lepus americanus, mientras que las aves llevadas a los nidos predominante- 
mente fueron Corvus brachyrhynchos y Bonasa umbellus. En promedio, los individuos nidificantes entre- 
garon 2.12 presas/d, y cada entrega tuvo un promedio de 275 g, para un total de 551 g entregados/d. 
Sin embargo, las tasas diarias {P < 0.001) y horarias {P = 0.01) de entrega de presas variaron entre 
nidos. Las tasas de entrega (P = 0.01) y la biomasa entregada (P = 0.038) incrementaron con el tamano 
de la nidada. La diversidad y equitabilidad de las presas consumidas fueron similares entre nidos y bajas 
a traves del area de estudio, probablemente debido a la dominancia de T. hudsonicus en la dieta. 

[Traduccion del equipo editorial] 


^ Present address and corresponding author: P.O. Box 1363, Meeker, CO 81641 U.S.A.; email: brett^mithers@blm.gov 


264 


September 2005 


Biology 


265 


The Northern Goshawk {Accipiter gentilis) is a 
large, forest-dwelling raptor generally associated 
with mature deciduous, coniferous, or mixed for- 
ests (e.g., Bright-Smith and Mannan 1994, Siders 
and Kennedy 1996, Beier and Drennan 1997, 
Squires and Reynolds 1997). Goshawk research in 
North America has been conducted primarily in 
the western half of the continent (Boal et al. 2003). 
Consequently, there is little published literature 
describing ecology of the species in the Western 
Great Lakes Region (WGLR) of North America, 
where it is currently listed as a Migratory Non- 
game Bird of Management Concern by the U.S. 
Fish and Wildlife Service (Region 3) and as a sen- 
sitive species by the U.S. Forest Service (Region 9) 
due to loss of habitat (Reynolds et al. 1992). 

Depending on region, season, and availability, 
goshawks capture a wide variety of prey and are 
considered prey generalists (Squires and Reynolds 
1997, Squires and Kennedy 2005). Although breed- 
ing-season diet composition has been studied for 
many populations (e.g., Meng 1959, Grzybowski 
and Eaton 1976, Boal and Mannan 1994, Younk 
and Bechard 1994, Lewis 2001), site-specific studies 
of diet are necessary for developing management 
strategies for goshawk populations at regional and 
local levels (e.g., Reynolds et al. 1992). A number 
of records exist of prey items collected opportu- 
nistically at goshawk nests in the WGLR (Eng and 
Gullion 1962, Apfelbaum and Haney 1984, Mar tell 
and Dick 1996), but these reports are anecdotal 
and provide a prey list rather than a quantitative 
assessment of food habits (Roberson et al. 2003) . 

Methods used in goshawk food habits research 
have included indirect (i.e., identification of prey 
remains or contents of regurgitated pellets) and 
direct observations of prey deliveries to nests 
(Meng 1959, Grzybowski and Eaton 1976, Bosa- 
kowski and Smith 1992, Boal and Mannan 1994). 
Indirect methods of assessing raptor diet can lead 
to biased results (e.g., Bielefeldt et al. 1992), 
whereas direct methods should provide the least- 
biased results (Collopy 1983, Marti 1987, Boal and 
Mannan 1994). During the breeding seasons of 
2000-02, we used videography as a modified meth- 
od of direct observation of prey deliveries to ex- 
amine diet of Northern Goshawks in northern 
Minnesota. 

Methods 

Study Area. The study area was located in the Lauren- 
tian Mixed-Forest Province of north-central and north- 


eastern Minnesota (46°50'N, 92°11'W) as described by 
Boal et al. (2001) and Roberson (2001; Fig. 1). The study 
area elevation ranged from ca. 200-400 m. Mean summer 
and winter temperatures were 18°C and — 11°C, respec- 
tively, and maximum and minimum temperature records 
for the region were 40°C and — 46°C, respectively (Daniel 
and Sullivan 1981). Annual precipitation averaged 60-70 
cm. The study area was dominated by pine, mixed-hard- 
wood, boreal, and second-growth forests with wetland 
community types interspersed among forest stands (Tes- 
ter 1995). 

Goshawk Nests. Nests included in this study were con- 
sidered as sampling units and were selected from all 
known occupied nests in the study area (Boal et al. 2001) . 
With the exception of one nest, where few data were col- 
lected during 2000, diet information was not collected at 
any nest for more than one breeding season. Nests were 
selected randomly within the constraints of accessibility 
and to include different land ownerships. Thus, our sam- 
ple is not truly random and may not be representative of 
the goshawk population of our study area. However, to 
examine the applicability of our diet data to the goshawk 
population as a whole, we examined prey diversity and 
overlap among nests. High overlap and low diversity 
would suggest prey use was similar among goshawk pairs 
and that our data were representative of the population 
in general. 

Video Recording. We used VHS (Model SL 800, Se- 
curity Labs®, Noblesville, IN U.S.A.) and 8-mm video re- 
cording systems (Sony® Model M-350, Fuhrman Diversi- 
fied, Inc., Seabrook, TX U.S.A.) with color or 
black-and-white cameras (Model CCM-660W, Clover Elec- 
tronics®, Los Alamitos, CA U.S.A.). Cameras were in- 
stalled on nest trees within 0.6 m of the nest or, for cam- 
eras with zoom lenses, on an adjacent tree up to 9 m 
from the nest. Video recorders were placed in weather- 
proof cases ca. 30 m from the base of each camera tree. 
Coaxial-video cables were used to convey power to and 
transmit images from the cameras. Recorders were pro- 
grammed to record from 0530-2100 H (15.5 hr of foot- 
age) at the 48-hr (1.3 frames/sec) or the 72-hr (0 8 
frames/sec) setting to optimize the amount of tape used 
per sampling session and battery life. We replaced tapes 
and batteries every 3-4 d. 

Prey Identification. To identify prey delivered to nests, 
we reviewed video footage until a prey delivery occurred, 
then advanced frame by frame and freeze-framed to fa- 
cilitate prey identification. We identified avian and mam- 
malian prey by morphological features and developed a 
list of prey species delivered by goshawks to all nests (Ta- 
ble 1). Goshawks may cache prey and retrieve cached 
prey items (Boal and Mannan 1994), which could bias 
estimates of delivery rates and proportional use of species 
in the diets. We attempted to identify cached prey on 
basis of a successive, iterative process that included com- 
paring prey items using flesh color, pelage or feather con- 
dition, and time of delivery from review of video footage, 
and then remove those items thought to be cached from 
analysis. 

Age and Biomass Estimation. We assigned avian prey 
to age categories (e.g., adult, juvenile, or nestling) based 
on plumage (e.g., feathers and down) and amount of 
sheathing on flight feathers (Reynolds and Meslow 


266 


Smithers et al. 


VoL. 39, No. 3 



Figure 1. Study area and distribution of Northern Goshawk nests in Minnesota where food habits information was 
collected during the 2000-02 breeding seasons. The three-letter designations indicate individual nests. Breeding 
season diet information collected at the DTK breeding area was omitted from all analyses because of nest failure. 
Breeding areas with similar prey composition are indicated with the same superscripts. Superscripts indicate cluster 
number (see Fig. 2). 


1984). We categorized mammalian prey as adults or ju- 
veniles based on size (Bielefeldt et al. 1992). Because of 
difficulty in estimating age of small mammals, we consid- 
ered all mammals smaller than chipmunks to be adults. 
Biomass for partial prey items was calculated using the 
proportion of prey delivered to nests, and proportions 
were estimated qualitatively (e.g., 50% of adult size). 

We estimated biomass for prey identified to family, ge- 
nus, or species and used the mean mass of both sexes 
(Reynolds and Meslow 1984, Lewis 2001). Biomass esti- 


mates were based on published information on mam- 
malian and avian species occurring in the study area 
(Burt and Grossenheider 1980, Jones and Birney 1988, 
Dunning 1993, Dunn and Garrett 1997, Dunn 1999, Sib- 
ley 2000). We calculated mass for nestlings following 
Bielefeldt et al. (1992) using 100% of the adult mass for 
warbler-sized species, 65% of the adult mass for robin 
and jay-sized species, and 55% of the adult mass for large 
birds such as grouse. We calculated mass of juvenile 
red squirrel ( Tamiasciurus hudsonicus ) , eastern chipmunk 


September 2005 


Biology 


267 


Table 1. Number, percent occurrence, and biomass of mammalian and avian prey delivered to Northern Goshawk 
nests {N = 13) in Minnesota, 2000—02. Values represent pooled number of prey identified at nests during the 2000, 
2001, and 2002 breeding seasons. 


Prey Category 

Common Name 

N 

Percent 

Biomass 

(g) 

Percent 

Mammals 






Tamiasdurus hudsonicus 

red squirrel 

202 

31.0 

38046 

23.6 

Tamias striatus 

eastern chipmunk 

95 

14.6 

8108 

5.0 

Lepus americanus 

snowshoe hare 

31 

4.8 

41027 

25.5 

Sylvilagus floridanus 

eastern cottontail 

7 

1.1 

7654 

4.8 

Sdurus carolinensis 

eastern gray squirrel 

3 

0.5 

1679 

1.0 

Peromyscus spp. 


2 

0.3 

47 

0.0 

Family: Muridae 


1 

0.2 

18 

0.0 

Mustela frenata 

long-tailed weasel 

1 

0.2 

210 

0.1 

Unknown mammal (MSC1)“ 


8 

1.2 

186 

0.1 

Unknown mammal (MSC2)® 


9 

1.4 

1720 

1.1 

Birds 






Corvus brachyrhynchos 

American Crow 

37 

5.7 

14515 

9.0 

Bonasa umbellus 

Ruffed Grouse 

33 

5.1 

18448 

11.5 

Ay thy a spp. 

diving duck 

12 

1.8 

11360 

7.1 

Cyanodtta cristata 

Blue Jay 

8 

1.2 

664 

0.4 

Fulica americana 

American Coot 

6 

0.9 

3338 

2.1 

Turdus migratorius 

American Robin 

3 

0.5 

205 

0.1 

Quiscalus quiscula 

Common Crackle 

3 

0.5 

341 

0.2 

Family: Icteridae 

blackbird 

3 

0.5 

189 

0.1 

Picoides spp. 

woodpecker 

3 

0.5 

199 

0.1 

Dryocopus pileatus 

Pileated Woodpecker 

3 

0.5 

861 

0.5 

Unknown duckling 


4 

0.6 

400 

0.2 

Butorides virescens 

Green Heron 

2 

0.3 

420 

0.3 

Perisoreus canadensis 

Gray Jay 

2 

0.3 

142 

0.1 

Agelaius phoeniceus 

Red-winged Blackbird 

2 

0.3 

105 

0.1 

Strix varia 

Barred Owl 

1 

0.2 

394 

0.2 

Buteo platypterus 

Broad-winged Hawk 

1 

0.2 

455 

0.3 

Genus: Calidris 


1 

0.2 

73 

0.0 

Bucephala clangula 

Common Goldeneye 

1 

0.2 

900 

0.6 

Accipiter cooperii 

Cooper’s Hawk 

1 

0.2 

439 

0.3 

Gallus spp. 

domestic chicken*’ 

1 

0.2 



Coccothraustes vespertinus 

Evening Grosbeak 

1 

0.2 

59 

0.0 

Pipilo erythrophthalmus 

Eastern Towhee 

1 

0.2 

41 

0.0 

Genus: Euphagus 


1 

0.2 

63 

0.0 

Acdpiter gentilis 

Northern Goshawk 

1 

0.2 

820 

0.5 

Picoides villosus 

Hairy Woodpecker 

1 

0.2 

66 

0.0 

Charadrius vodferus 

Killdeer 

1 

0.2 

97 

0.1 

Anas platyrhynchos 

Mallard 

1 

0.2 

1082 

0.7 

Sitta canadensis 

Red-breasted Nuthatch 

1 

0.2 

10 

0.0 

Sdurus aurocapillus 

Ovenbird 

1 

0.2 

19 

0.0 

Catharus fuscescens 

Veery 

1 

0.2 

31 

0.0 

Unknown nestling 


33 

5.1 

1190 

0.7 

Unknown bird (ASCl)^ 


18 

2.8 

173 

0.1 

Unknown bird (ASC2)® 


23 

3.5 

1778 

1.1 

Unknown bird (ASC3)^ 


6 

0.9 

3459 

2.1 

Items not identified to class 






Mammalia or Aves 


76 

11.7 



■* MSCl = mouse-sized prey item; MSC2 

= red squirrel-sized prey item; ASCI = 

warbler-sized prey 

item: ASC2 = 

robin-sized prey 


Item; ASC3 = Ruffed Grouse-sized prey item. 
^ Omitted from analysis. 


268 


Smithers et al. 


VoL. 39, No. 3 


{Tamias striatus), snowshoe hare {Lepus americanus), and 
eastern cottontail (Sylvilagus floridanus) using 95% of the 
adult mass; if ages could not be determined reliably, we 
assigned juvenile masses to these species. 

To estimate biomass of unidentified prey, we pooled 
unidentified birds into three a priori size classes (SC) fol- 
lowing Storer (1966) and Kennedy and Johnson (1986) 
that represented average mass of common species in our 
study area: SCI = 10 g (e.g., warbler-sized), SC2 = 77 g 
(e.g., robin-sized), and SC3 = 576 g (e.g., Ruffed Grouse 
[Bonasa umbellus]-sized) , Similarly, we pooled unidenti- 
fied mammal prey into two a priori size classes: SCI = 23 
g (e.g., mouse-sized) and SC2 = 192 g (e.g., squirrel- 
sized) . 

Prey and Biomass Delivery Rates. We calculated deliv- 
ery rates on the basis of number of prey delivered per 
day, number of prey delivered per nestling per day, and 
number of prey delivered per day at nests with one, two, 
and three nestlings. We calculated biomass estimates in 
the same manner. We calculated mean delivery rates over 
5-d intervals from hatching to 5 d post-fledging (i.e., 
from 0-45 d). 

Prey Diversity and Overlap. We calculated prey diver- 
sity for the study area using ungrouped prey categories 
(i.e , using each prey category identified to family, genus, 
or species separately) . Because samples were smaller 
when examining individual nests, we generalized prey 
into similar species categories (Lewis 2001) to calculate 
prey diversity for individual nests. The generalized prey 
categories for among-nest diversity assessment were: (1) 
Sciurids, (2) blackbirds and Corvids, (3) Leporids, (4) 
Ruffed Grouse, (5) diving ducks (Aythya spp.) , (6) water 
and shore birds, (7) passerines, (8) Picidae, (9) Falcon- 
iforms, (10) miscellaneous mammals (e.g., long-tailed 
weasel [Mustela frenata ] ) . 

We calculated prey diversity using Williams (1964) and 
MacArthur’s (1972) modified form of the Simpson’s in- 
dex (Simpson 1949) and diet equitability using Smith 
and Wilson’s index of evenness (Smith and Wilson 1996) . 
We used prey identified to family, genus, or species to 
estimate diet overlap among nests with the Simplified 
Morisita’s Index of Overlap (Krebs 1999). Overlap mea- 
sures are designed to measure the degree that two spe- 
cies share a set of common resources or utilize the same 
parts of the environment (Lawlor 1980). Overlap mea- 
sures are scaled from zero to one, where zero overlap 
indicates dissimilarity in resource use, and one indicates 
complete overlap (Krebs 1999). We also assessed similar- 
ity in prey use among nests with cluster analysis using 
average linkage clustering (Romesburg 1984, Krebs 1999, 
McGarigal et al. 2000). As suggested by Romesburg 
(1984), we used the un-weighted pair-group method us- 
ing arithmetic averages (UPGMA). 

Statistical Analysis. We used analysis of variance (AN- 
OVA) to examine relationships between delivery rate var- 
iables and brood size using log-transformed data (Zar 
1999). Biomass of prey delivered per day per nest was 
transformed by taking the logarithm of biomass delivered 
per day and adding 1.0 (Zar 1999). Normality of exper- 
imental error was tested using the Shapiro-Wilk test pro- 
cedure, and assumptions regarding homogenous varianc- 
es were tested using Levene’s test (Zar 1999). We 
examined differences in the number of mammals and 


birds delivered among nests over 5-d intervals, because 
of missing data among sampled days, with a Kruskal-Wal- 
lis single-factor ANOVA (Zar 1999). Because observations 
within breeding areas were not independent, we exam- 
ined differences in provisioning rates among breeding 
areas with multivariate repeated measures ANOVA. We 
used the General Linear Model (GLM) module of STA- 
TISTICA (Version 6.0, StatSoft, Inc., Tulsa, OK U.S.A.) 
for all statistical analyses except calculation of diet over- 
lap and similarity, for which we used Ecological Meth- 
odology 6.1 (Exeter Software, Setauket, NYU.S.A.). An 
alpha level of P = 0.05 was used for all statistical tests, 
and we present means and standard errors. 

Results 

Video Recording and Prey Identification. We in- 
stalled video monitoring systems at three, five, and 
seven occupied goshawk nests during the 2000, 
2001, and 2002 field seasons, respectively. We 
placed cameras at nests when nestlings were ca. 8 
d old (±1.18; range = 1-18 d). One of the 15 nests 
failed within 3 d of camera placement and was re- 
moved from analysis. Due to camera malfunctions, 
we were only able to collect 16 hr of footage at one 
of the nests in 2000. We placed a camera at the 
2002 nest of the same pair, but pooled data from 
both years as one nest area for analysis. Thus, our 
sample of 4801 hr (x = 320 ± 42 hr/nest) of video 
footage is derived from 13 nesting pairs of gos- 
hawks. 

We identified 59 (8.3%) of 711 prey deliveries as 
being retrievals of cached items. Of the 652 fresh 
prey deliveries, we identified 451 (69%) to the spe- 
cies level, 20 (3%) to genus, four to family (1%), 
and four (1%) as unidentifiable ducklings (Table 
1). Eighty (12%) birds and 17 (3%) mammals were 
unidentifiable beyond class, and we were unable to 
identify 76 (12%) deliveries. The majority of prey 
deliveries identified to at least class {N = 576) were 
mammals (62%), whereas birds (38%) comprised 
a smaller proportion of diet. 

When considering only those deliveries identi- 
fied to family or finer resolution (i.e., to genus or 
species; N = 476), the dominant prey species were 
red squirrels (41.2%), eastern chipmunks (19.8%), 
American crows (7.7%), Ruffed Grouse (6.9%), 
and snowshoe hares (6.5%). No other individual 
species accounted for >5% of identified prey. As a 
group, Sciurids and Leporids {N = 338) accounted 
for 70% of the identified prey. Among mammals, 
51.8% were adults, 25.4% were Juveniles, and we 
were unable to estimate age for 22.8%. Of the 
birds, 36.7% were adults, 9.6% were Juveniles, 


September 2005 


Biology 


269 


27 . 5 % were nestlings, and we could not reliably 
estimate age for 26 . 2 %. 

Biomass. In context of the prey species and bio- 
mass proportion used by goshawks in our study, the 
delivery of one domestic chicken {Gallus spp.) was 
unusual and the mass would dramatically influence 
biomass estimates for avian prey. We therefore con- 
sidered it an outlier and deleted it from biomass 
estimates. 

We estimated the total biomass of all prey deliv- 
eries at nests as 161 kg. The mean mass for both 
avian and mammalian prey was 281 g (± 13 . 7 , 95 % 
confidence interval = 254—308 g). Although aver- 
age mass of avian prey {x = 292 g; range = 10 - 
1082 g) was similar to that for mammalian prey 
(275 g; range = 18-1361 g), avian prey accounted 
for only 39 % of biomass delivered whereas mam- 
mals accounted for 61 % of biomass delivered. 
Snowshoe hare ( 25 %), red squirrel ( 24 %), Ruffed 
Grouse ( 11 %), American Crow ( 9 %), diving ducks 
( 7 %), chipmunk ( 5 %), and eastern cottontail 
( 5 %) accounted for 86 % of biomass used by gos- 
hawks. No other species accounted for >5% of bio- 
mass. 

Delivery Rates. Breeding goshawks delivered 
2.12 (± 0 . 14 ) prey per day (i.e., 0.14 deliveries/hr), 
each delivery had a mean mass of 275 g (±20 g), 
for a total of 551 g (±50 g) delivered per day. How- 
ever, daily ( 7 ^ 13,253 = 3 . 44 , P < 0 . 001 ) and hourly 
( 7 ^ 13,250 = 2 . 31 , P = 0 . 01 ) delivery rates varied 
among nests. 

1.3 (± 0 . 1 ) prey items were delivered per nestling 
per day, but delivery rates increased with brood 
size ( 7 ^ 2,271 “ 5 . 23 , P = 0 . 01 ). Daily prey delivery 
rates were 1.8 (± 0 . 1 ) at nests with one nestling, 2.3 
(± 0 . 1 ) at nests with two nestlings, and 2.5 (± 0 . 2 ) 
at nests with three nestlings. Despite the increase 
in prey deliveries among nests with larger broods, 
there was an inverse relationship between brood 
size and the number of prey delivered per nestling 
per day (r = — 0 . 43 , P < 0 . 05 ). Each nestling in 
single broods received a mean of 1.8 (± 0 . 1 ) prey 
items per day, whereas each nestling in broods of 
two received only 1.2 (± 0 . 1 ) prey items per day, 
and each nestling in broods of three received only 
0.9 (± 0 . 1 ) prey items per day. 

322 g (±32 g) of biomass were delivered per 
nestling. However, we observed a pattern of bio- 
mass delivered to broods of different sizes that was 
similar to that of number of prey delivered to 
broods of different sizes; biomass delivered per 
nestling per day (Eg 5 = 5 . 96 , P = 0 . 038 ) varied with 


brood size. On average, daily biomass delivered was 
509 g (±84 g) to nests with one nestling, 555 g 
(±42 g) to broods of two, and 756 g (±107 g) to 
broods of three. Despite greater amounts of bio- 
mass being provided to larger broods, this resulted 
in nestlings in single broods receiving 509 g (±84 
g) of biomass per day, whereas nestlings in broods 
of two each received 278 g (±3 g) of biomass per 
day and nestlings in broods of three each receiving 
252 g (±36 g) per day. 

Dietary Overlap. The diversity and equitability of 
prey delivered to nests was low for the study area, 
as indicated by a reciprocal of the Simpson diver- 
sity index ( 1 / 7 )) of 4.28 and a Smith and Wilson 
evenness index (i^ar) of 0 . 30 . Similarly, diversity 
among nests was low, with a mean value of l/D = 
3.77 (± 0 . 41 , range = 2 - 09 - 7 . 35 ). The mean value 
of for all nests was 0.56 (± 0 . 04 , range = 0 . 36 - 
0 . 80 ). Low prey diversity and evenness values may 
be attributable to goshawk diet being dominated 
by red squirrels and chipmunks in our study. Sim- 
ilarly, there was high dietary overlap (> 0 . 8 ) among 
breeding pairs of goshawks in our study (Table 2 ) , 
although one nesting area (LSP; Table 2 ) ap- 
peared to be measurably different from the rest. 
Cluster analysis indicated there were two groups of 
breeding goshawk diets that exhibited similar prey 
composition and proportion of use (Fig. 2 ) al- 
though, again, one nest (LSP; Fig. 2 ) appears to 
be an outlier. There was no apparent relationship 
between overlap measures and spatial distribution 
of nests across the study area (Fig. 1 , 2 ). 

Discussion 

Mammals were the dominant prey of breeding 
goshawks in Minnesota, with red squirrels and east- 
ern chipmunks appearing to be the most impor- 
tant species in terms of both number delivered and 
biomass. These two species alone accounted for 
62 % of all prey identified to at least family and 
51 % of prey identified to at least class. Several stud- 
ies have documented red squirrels as important 
prey for goshawks (Squires and Kennedy 2005 ) 
throughout their range. They may be especially im- 
portant during the winter when other prey may be 
less available (Widen et al. 1987 ). Squirrels domi- 
nated goshawk diets in Sweden in terms of number 
( 79 %) and biomass ( 56 %) during winters of both 
high and low squirrel abundance (Widen et al. 
1987 ). Diet information for winter goshawks in the 
WGLR is not available, but the extensive use of red 
squirrels during the summer and the patterns of 


Table 2. Dietary overlap values using the Simplified Morisita’s Index of Overlap. Values range from 0 (no overlap) to 1 (complete overlap). The data presented 
were generated from prey frequency data collected at Northern Goshawk nests {N = 13) in Minnesota during the 2000, 2001, and 2002 breeding seasons. The 
three letter codes designate specific goshawk nests. 


VOL. 39, No. 3 


270 


Smithers et al. 




CO 



05 

xf5 

CO 


so 

t- 

xf5 

J> rH 




00 

xO 


00 

’sh 

rH 

05 

r-H 


XO 


i> 

O] 

05 

05 

xO 

00 

05 

05 

05 

05 

00 

00 


o 

d 

d 

d 

d 

d 

d 

d 

d 

d 

d 

d 

< 

00 


05 

I-H 

00 

00 

XO 




so 




05 


CM 

00 

xn 

00 

cO 

cO 

05 



cn 

00 

00 

CJ5 

cO 

05 

05 

05 

00 

05 

00 



d 

d 

d 

d 

d 

d 

d 

d 

d 

d 

d 





CM 

05 

I— c 

(M 


CM 

05 

05 

^H 



50 

o 

so 

r- 

00 

o 

rH 

CO 

00 




00 

00 

05 

05 

r- 

00 

CJ5 

05 

00 

00 




d 

d 

d 

d 

d 

d 

d 

d 

d 

d 






00 

so 





00 

r- C 



xn 

o 

so 

IC 

so 

C05 

00 

05 

CM 




H 


CO 

05 

05 

CO 

C05 

05 

05 

05 




(Z) 

d 

d 

d 

d 

d 

d 

d 

d 

d 





i> 

00 

05 

05 


so 

CO 

CM 

f-H 






00 

00 

xO 

CO 

05 

XO 

SO 





|*q 


a> 

05 

05 

xq 

00 

C05 

05 





03 

d 

d 

d 

d 

d 

d 

d 

d 





H 


m 

00 

05 

05 

so 

CJ5 

I-H 





ID 

o 


CO 

CM 

05 

00 






s 

o> 

05 

05 

05 

CO 

05 

05 






Cu 

d 

d 

d 

d 

d 

d 

d 






Q 

U 




05 

05 

so 







CM 

so 


J> 

so 

r-- 







<y> 

05 

05 

05 

cO 

05 








d 

d 

d 

d 

d 

d 







<1 

tJD 




CM 

r*H 







r- 


T-^ 


CM 








(Z) 

cn 

00 

05 

05 

CD 








hJ 

d 

d 

d 

d 

d 









to 


CM 

CM 










CO 

CM 

05 

so 









(Z) 

CO 

xq 

xO 

CO 










d 

d 

d 

d 










CO 

CM 

00 

r-f 









w 

>-n 

00 

XO 

00 










00 

05 

05 











d 

d 

d 










o 

vr> 

CO 

00 

I— 1 











00 

05 












d 

d 











X 


I-H 
























Q 

d 












w 













w 













p 














DEE 

DIX 

HAG 

z 

w 

LSP 

LSA 

MCD 

PMT 

SHA 

STE 

WAG 

WFA 

WRI 


squirrel use during winter in other areas (Widen 
et al. 1987) suggest this species may be of year- 
round importance to goshawks in the region. In 
terms of biomass, snowshoe hares also appear to 
be important for goshawks in our study area, ac- 
counting for 25% of the biomass delivered to nests. 
Rabbits and hares are also used extensively by gos- 
hawks throughout their range (Squires and Ken- 
nedy 2005). 

Ruffed Grouse comprised 5% of prey deliveries 
and 11% of biomass delivered to goshawk nests 
during a 3-yr period of relatively low grouse abun- 
dance (Smithers 2003). There is anecdotal evi- 
dence that at least some goshawks in Minnesota 
may rely more heavily on Ruffed Grouse than oth- 
er prey during some time periods (Eng and Gul- 
lion 1962, Apfelbaum and Haney 1984). Eng and 
Gullion (1962) focused on Ruffed Grouse mortal- 
ity and did not assess proportional use of grouse 
in the diet of goshawks, and Apfelbaum and Haney 
(1984) reported on prey remains collected at a sin- 
gle nest in northern Minnesota. Because of the dif- 
ficulties in accurately quantifying the extent of 
grouse predation by goshawks (Eng and Gullion 
1962) and the biases associated with determining 
raptor diets based on prey remains (Smithers 
2003), the results of these studies need to be in- 
terpreted cautiously. We suspect that the previous 
research on goshawk diet for our study area, all 
collected by indirect methods (Eng and Gullion 
1962, Apfelbaum and Haney 1984, Martell and 
Dick 1996), may overestimate the proportion of 
birds, especially large birds such as grouse, and un- 
derestimate the proportion of mammals in gos- 
hawk diets. 

Qualitative review of the data suggests the mean 
delivery rate of 0.14 deliveries/hr to nests in our 
study was less than that observed in Arizona (0.25 
deliveries/hr; Boal and Mannan 1994), Nevada 
(0.31 deliveries/hr; Younk and Bechard 1994) and 
two areas of southeast Alaska (0.30 and 0.23 deliv- 
eries/hr; Lewis 2001). However, although mean 
biomass per delivery in our study (275 g) was less 
than that in Arizona (307 g/delivery) where Le- 
porids and Sciurids were the dominant prey (Boal 
and Mannan 1994), it was greater than the two ar- 
eas of Alaska (214 g and 173 g/delivery), where 
birds were the dominant prey (Lewis 2001). 

Our study indicates that goshawks with larger 
broods provision with greater delivery rates and 
biomass. Biomass per nestling was similar between 
broods of two and three (16.3-18.0 g/hr), but only 


September 2005 


Biology 


271 



Figure 2. Cluster analysis dendrogram for food habits data collected at Northern Goshawk nests in Minnesota during 
the 2000, 2001, and 2002 breeding seasons. Parentheses indicate cluster number (see Fig. 1). The LSP and WAG 
breeding areas exhibited the least similarity of diet composition among breeding areas. 


about half as much as that received by nestlings in 
broods of one (33.0 g/hr). This poses an interest- 
ing question regarding energetic aspects of gos- 
hawk productivity; what is the minimum biomass/ 
hr necessary to fledge young successfully? The 
similarity between broods of two and broods of 
three suggests that, at least in our study area, and 
at nests with similar prey composition, a minimum 
of 16—18 g of biomass per hr may be required for 
successful nesting. However, a finer assessment of 
nestling energetics would likely require experi- 
mentation in a laboratory setting. 

Given our prey use and delivery rate data, one 
can make a generalized prediction of the relative 
impact of a breeding pair of goshawks in our study 
area during the 45-d nestling period. With an ex- 
pected delivery rate of 2.1 prey/ d over a 45-d nest- 
ling period, ca. 94 prey deliveries can be expected. 


Based on observed frequencies of prey use, this 
would translate to the average breeding goshawk 
pair capturing 29 red squirrels, 14 eastern chip- 
munks, six American Crows, five snowshoe hares, 
five Ruffed Grouse, two diving ducks, one cotton- 
tail, one Blue Jay, and 31 miscellaneous small birds 
and mammals. To put this level of predation in 
context, all of these prey captures would occur 
within a home range of 6376 ha for a goshawk pair 
in the study area (Boal et al. 2003) . 

Composition and richness of prey delivered to 
nests was similar across the study area, and esti- 
mates of prey diversity and equitability were gen- 
erally low among nests. We suspect the high dietary 
overlap and similarity of prey use among breeding 
areas was most likely attributable to the dominance 
of red squirrels and chipmunks in goshawk diets. 
However, goshawk diets were dominated by red 


272 


Smithers et al. 


VoL. 39, No. 3 


squirrels and chipmunks, but snowshoe hare, 
Ruffed Grouse, and American Crow were also im- 
portant in terms of biomass. 

As pointed out by Reynolds et al. (1992), raptor 
populations are often limited by prey availability 
and their choice of foraging habitat is predicated 
on conditions in which prey are abundant and 
available. Thus, an understanding of goshawk prey 
species used and the relative importance of those 
prey species is an important step toward develop- 
ing management plans for goshawks. By identifying 
key prey species, as we have done here, forest man- 
agers can develop a set of desirable conditions that 
fosters presence of those species while incorporat- 
ing structural aspects of known goshawk foraging 
habitat (e.g., Boal et al. 2001). Those desirable for- 
est conditions can be incorporated into goshawk 
management plans as one factor of foraging habi- 
tat (e.g., Reynolds et al. 1992) and facilitate con- 
servation of the species. 

Acknowledgments 

A. Bellman, F. Nicoletti, J. Ridelbauer, A. Roberson, M. 
Solensky, L. Smithers, W. Steffans, C. Trembath, and A. 
Wester assisted with various aspects of this study. Person- 
nel from the many cooperating agencies and organiza- 
tions provided assistance and logistical support during 
this project. They included R. Baker, J. Casson, J. Gal- 
lagher, L. Grover, M. Hamady, J. Hines, M. Houser, E. 
Lindquist, C. Mortensen, S. Mortensen, B. Ohlander, W. 
Russ, R. Vora, and A. Williamson. Video equipment for 
the 2000 and 2001 field seasons was provided by Alaska 
Department of Fish and Game and the Wisconsin De- 
partment of Natural Resources. Funding and logistical 
support for this project was provided by the Minnesota 
Department of Natural Resources, Minnesota Forest In- 
dustries, The National Council for Air and Steam Im- 
provement, Potlatch Corporation, the U.S. Forest Service 
Chippewa and Superior National Forests, and the U.S. 
Fish and Wildlife Service. 

Literature Cited 

Apfelbaum, S.I. and A. Haney. 1984. Note on foraging 
and nesting habitats of goshawks. Loon 56:132-133. 
Beier, P. and J.E. Drennan. 1997. Forest structure and 
prey abundance in foraging areas of Northern Gos- 
hawks. Ecol. Appl. 7:564—571. 

Bielefeldt, J., R.N. Rosenfield, and J.M. Papp. 1992. Un- 
founded assumptions about diet of the Cooper’s 
Hawk. Condor 94:427-436. 

Boal, C.W., D.E. Andersen, and P.L. Kennedy. 2001. 
Home range and habitat use of Northern Goshawks 
in Minnesota. Final Report. Minnesota Cooperative 
Fish and Wildlife Research Unit, University of Min- 
nesota, St. Paul, MN U.S.A. 

, , AND . 2003. Home range and res- 


idency status of Northern Goshawks breeding in Min- 
nesota. Condor 105:811-816. 

AND R.W. Mannan. 1994. Northern Goshawk diets 

in ponderosa pine forests on the Kaibab Plateau. Stud. 
Avian Biol. 16:97-102. 

Bosakowski, T. AND D.G. Smith. 1992. Comparative diets 
of sympatric nesting raptors in the eastern deciduous 
forest biome. Can. J. Zool. 70:984-992. 

Bright-Smith, D.J. and R.W. Mannan. 1994. Habitat use 
by breeding male Northern Goshawks in Northern 
Arizona. Stud. Avian Biol. 16:58—65. 

Burt, WH. and R.P. Grossenheider, 1980. A field guide 
to the mammals. 3rd Ed. Houghton Mifflin Company, 
Boston, MA U.S.A. 

COLLOPY, M.W. 1983. A comparison of direct observation 
and collections of prey remains in determining the 
diet of Golden Eagles./. Wildl. Manag. 47:360-368. 

Daniel, G. and Sullivan, J. 1981. A Sierra Club natural- 
ist’s guide to the north woods of Michigan, Wisconsin, 
Minnesota and southern Ontario. Sierra Club Books, 
San Franciso, CA U.S.A. 

Dunn, J.L. 1999. Field guide to the birds of North Amer- 
ica, 3rd Ed. National Geographic Society, Mary B. 
Dickinson, Ed., Washington, DC U.S.A. 

AND K. Garrett. 1997. A field guide to warblers 

of North America. Houghton Mifflin Company, Bos- 
ton, MA U.S.A. 

Dunning, J.B., Jr. 1993. CRC handbook of avian body 
masses. CRC Press, Inc., Boca Raton, FL U.S.A. 

Eng, R.L. and G.W. Gullion. 1962. The predation of 
goshawks upon Ruffed Grouse on the Cloquet Forest 
Research Center, Minnesota. Wilson Bull. 74:227-241. 

Grzybowski, J.A. AND S.W. Eaton. 1976. Prey items of 
goshawks in southwestern New York. Wilson Bull. 88: 
669-670. 

Jones, J.K., Jr. and E.C. Birney. 1988. Handbook of mam- 
mals of the north-central states. Univ. Minnesota 
Press, Minneapolis, MN U.S.A. 

Kennedy, P.L. and D.R. Johnson. 1986. Prey size selec- 
tion in nestling male and female Cooper’s Hawks. Wil- 
son Bull. 98:110-115. 

Krebs, C.J. 1999. Ecological methodology, 2nd Ed. Ad- 
dison-Wesley Educational Publishers, Inc., Reading, 
MA U.S.A. 

Lawlor, L.R. 1980. Overlap, similarity and competition 
coefficients. Ecology 61:245-251. 

Lewis, S.B. 2001. Breeding season diet of Northern Gos- 
hawks in southeast Alaska with a comparison of tech- 
niques used to examine raptor diet. M.S. thesis, Boise 
State Univ., Boise, ID U.S.A. 

McGarigal, K., S. Cushman, and S. Stafford. 2000. Mul- 
tivariate statistics for wildlife and ecology research. 
Springer, New York, NY U.S. A. 

MacArthur, R.H. 1972. Geographical ecology. Harper 
and Row, New York, NY U.S.A. 

Marti, C.D. 1987. Raptor food habits studies. Pages 67- 
80 in B.A. Giron Pendleton, B.A. Millsap, K.W. Cline, 


September 2005 


Biology 


273 


and D.M. Bird [Eds.], Raptor management tech- 
niques manual. National Wildlife Federation, Wash- 
ington, DC U.S.A. 

Martell, M. and T. Dick. 1996. Nesting habitat charac- 
teristics of the Northern Goshawk {Acdpiter gentilis) in 
Minnesota. Final Report to Minnesota Department of 
Natural Resources, Nongame Wildlife Program, Proj- 
ect No. 9407382., Minnesota Department of Natural 
Resources, St. Paul, MN U.S.A. 

Meng, H. 1959. Food habits of nesting Cooper’s Hawks 
and goshawks in New York and Pennsylvania. Wilson 
Bull. 71:169-174. 

Reynolds, R.T. and E.C. Meslow. 1984. Partitioning of 
food and niche characteristics of coexisting Acdpiter 
during breeding. Auk 101:761-779. 

, R.T. Graham, M.H. Reiser, R.L. Bassett, P.L. 

Kennedy, D.A. Boyce, Jr., G. Goodwin, R. Smith, and 
E.L. Fisher. 1992. Management recommendations for 
the Northern Goshawk in the southwestern United 
States. USDA Forest Service, Gen. Tech. Rep. RM-217, 
Ft. Collins, CO U.S.A. 

Roberson, A.M. 2001. Evaluating and developing survey 
techniques using broadcast conspecific calls for 
Northern Goshawks in Minnesota. M.S. thesis, Univ. 
Minnesota, St. Paul, MN U.S.A. 

, D.E. Andersen, and P.L. Kennedy. 2003. The 

Northern Goshawk {Acdpiter gentilis atricapillus) in the 
western Great Lakes region: a technical conservation 
assessment. Minnesota Cooperative Fish and Wildlife 
Research Unit, Univ. Minnesota, St. Paul, MN U.S.A. 

Romesburg, H.C. 1984. Cluster analysis for researchers. 
Lifetime Learning Publications, Belmont, CA U.S.A. 

Sibley, D.A. 2000. The Sibley guide to birds. National 
Audubon Society, Alfred A. Knopf, Inc., New York, NY 
U.S.A. 

SiDERS, M.S. and P.L. Kennedy. 1996. Forest structural 
characteristics of accipiter nesting habitat; is there an 
allometric relationship? ContJor 98:123-132. 


Simpson, E.H. 1949. Measurement of diversity. Nature 
163:688. 

Smith, B. and J.B. Wilson. 1996. A consumer’s guide to 
evenness indices. Oikos 76:70-82. 

Smithers, B.L. 2003. Northern Goshawk food habits in 
Minnesota: an analysis using time-lapse recording sys- 
tems. M.S. thesis, Texas Tech University, Lubbock, TX 
U.S.A. 

Squires, J.R. and R.T. Reynolds. 1997. Northern Gos- 
hawk {Acdpiter gentilis) . In A. Poole and F. Gill [Eds.], 
The birds of North America, No. 298. The Birds of 
North America Inc., Philadelphia, PA U.S.A. 

AND P.L. Kennedy 2005. Northern Goshawk ecol- 
ogy: an assessment of current knowledge and infor- 
mation needs for conservation and management. 
Stud. Avian Biol. 31:in press. 

StatSoft, Inc. 2001. STATISTICA; the small book. 
StatSoft, Inc., Tulsa, OK U.S.A. 

Stoker, R.W. 1966. Sexual dimorphism and food habits 
in three North American accipiters. Auk 83:423-246. 

Tester, J.R. 1995. Minnesota’s natural heritage: an eco- 
logical perspective. Univ. Minnesota, Minneapolis, 
MN U.S.A. 

Widen, R, H. Andren, P. Angelstam, and E. Lindstrom. 
1987. The effect of prey vulnerability: goshawk pre- 
dation and population fluctuations of small game. Oi- 
kos 49:233-235. 

Williams, C.B. 1964. Patterns in the balance of nature. 
Academic Press, London, U.K. 

Younk, J.V. and MJ. Bechard. 1994. Breeding ecology of 
the Northern Goshawk in high-elevation aspen forest 
of northern Nevada. Stud. Avian Biol. 16:119-121. 

Zar, J.H. 1999. Biostatistical analysis, 4th Ed. Prentice- 
Hall, Inc., Englewood Cliffs, NJ U.S.A. 

Received 22 March 2004; accepted 28 September 2004 

Guest Editor; Stephen DeStefano 


J. Raptor Res. 39(3):274-285 
© 2005 The Raptor Research Foundation, Inc. 


SAMPLING CONSIDERATIONS FOR DEMOGRAPHIC AND 
HABITAT STUDIES OF NORTHERN GOSHAWKS 

Richard T. Reynolds, ^ J. David Wiens, Suzanne M. Joy, and Susan R. Salafsky 

Rocky Mountain Research Station, 2150 Centre Avenue, Building A, Suite 350, Fort Collins, CO 80526-1891 U.S.A. 

Abstract. — ^We used mark-recapture methods to monitor Northern Goshawks {Accipiter gentilis) and 
their nests over 12 yr in an increasing sample of breeding territories (37 in 1991 to 121 in 2002) in 
northern Arizona. As many as 8 yr of repeated nest searching were required to identify the population 
of breeders, as individuals skipped egg-laying on territories for up to 7 consecutive yr. Extensive temporal 
(within territory) and spatial (among territory) variation in reproduction and a high annual frequency 
of movements among dispersed alternate nests in territories made finding and monitoring goshawks 
problematic. Low detectability of nonbreeding goshawks (combined with uncertainties stemming from 
variations in breeding and use of alternate nests) made it difficult to categorize territories unequivocally 
as “unoccupied” by goshawks in non-egg-laying years. Temporal and spatial variation in reproduction 
required large numbers of territories to attain reliable estimates of reproduction of goshawks; such 
estimates were achieved only when samples approached or exceeded 60—100 territories. Our within- 
territory goshawk searching protocol, designed to increase the low and variable detectability of goshawks, 
required extensive sampling efforts to detect among-alternate nests movements. In lieu of such efforts, 
samples of territories occupied by goshawks may “decay” over time and lead to false inferences of 
population declines. Low detectability, variations in breeding, and large samples require that demo- 
graphic and habitat studies of goshawks employ intensive and repeated searches for goshawks in large 
study areas over at least 8 yr. 

Key Words: Northern Goshawk, Accipiter gentilis; alternate nest, Arizona; bootstrap; detectability; monitoring, 

reproduction; sampling, territory occupancy. 


GONSIDERACIONES SOBRE EL MUESTREO EN ESTUDIOS DEMOGRAFICOS Y DE HABITAT DE 
ACCIPITER GENTILIS 

Resumen. — Usamos tecnicas de captura-recaptura para evaluar las actividades de Accipiter gentilis y dc sus 
nidos a lo largo de 12 ahos en una muestra creciente de territorios de nidificacion (37 en 1991 a 121 
en 2002) en el norte de Arizona. Para poder identificar la poblacion reproductiva de Accipiter gentilis, 
requerimos hasta 8 ahos de busqueda repetida de nidos, ya que esta especie evito poner huevos en 
territorios por periodos de hasta 7 ahos consecutivos. La gran variacion temporal (dentro de los terri- 
torios) y espacial (entre territorios) en la reproduccion y una alta frecuencia anual de movimientos 
entre nidos alternos disperses en los territorios dificulto encontrar y evaluar las actividades de A. gentilis. 
La baja detectabilidad de los individuos no-reproductivos de A. gentilis (combinado con incertidumbres 
provenientes de las variaciones en la reproduccion y en el uso alterno de nidos) hizo dificil categorizar 
los territorios de modo inequivoco como “desocupados” por A. gentilis en los ahos en que no pusieron 
huevos. Esta dificultad se manifesto por la presencia de nidos activos de los mismos individuos anillados 
de A. gentilis luego de mas de un aho sin presencia reproductiva en los territorios. La variacion temporal 
y espacial en la reproduccion requirio grandes numeros de territorios para alcanzar estimaciones con- 
fiables de reproduccion en A. gentilis. Estas estimaciones fueron obtenidas solo cuando las muestras 
alcanzaron o excedieron los 60-100 territorios. Nuestro protocolo de busqueda de A. gentilis denXxo de 
los territorios, disehado para incrementar la detectabilidad baja y variable de A. gentilis, requirio es- 
fuerzos de muestreo amplios para detectar movimientos entre nidos alternos. Sin estos esfuerzos, las 
muestras de los territorios ocupados por A. gentilis podrian “disminuir” a lo largo del tiempo y llevarnos 
a inferencias falsas sobre disminuciones poblacionales. La baja detectabilidad, las variaciones reprod- 
uctivas y la necesidad de muestras de gran tamaho requieren que los estudios demograficos y de habitat 


^ E-mail address: rreynolds@fs.fed.us 


274 


September 2005 


Techniques 


275 


de A. gentilis empleen busquedas intensivas y repetidas de esta especie en grandes areas de estudio 
durante al menos ocho anos. 

[Traduccion del equipo editorial] 


The distribution, abundance, vital rates, and 
habitat occupancy of Northern Goshawks (Accipiter 
gentilis) are difficult to determine because of their 
elusive behavior in structurally-complex habitats, 
their low breeding densities, and annually variable 
breeding rates (DeStefano et al. 1994, Reynolds et 
al. 1994, Kennedy 1997, Reynolds and Joy in press). 
While locating and monitoring nests are common 
approaches in studies of avian demography and 
habitat, making valid inferences to a target popu- 
lation depends on reliable (unbiased and precise) 
estimates of the distribution and abundance of 
nests, demographic rates at nests, and habitat oc- 
cupancy. In such studies, it is often too costly to 
detect all individuals and to sample all areas, mak- 
ing a census (complete count) impractical, espe- 
cially in difficult-to-detect species. For such species, 
population parameters and habitat occupancy are 
often estimated using sampling methods. Making 
inferences about a species’ distribution or habitat 
occupancy from samples requires inferences about 
the species’ detection probability (probability that 
an individual is included in a sample when pres- 
ent). Biologists attempt to minimize influences of 
incomplete observations on estimates of a species’ 
distribution, demographics, and habitats with sam- 
pling frameworks that increase the detection rates 
of the species (Peterson and Bayley 2004, Mc- 
Donald 2004) . The problem is to understand how 
detectability varies within and among individuals, 
both temporally (year-to-year) and spatially 
(among territories) , and to develop sampling pro- 
tocols and efforts that increase detection rates of 
all individuals. 

We used mark-recapture methods from 1991- 
2002 to determine the distribution, abundance, vi- 
tal rates, fidelity to mate and territory, natal and 
breeding dispersal, and habitat occupancy of gos- 
hawks breeding on territories that increased in 
number from 37 in 1991 to 121 in 2002 (Reynolds 
et al. 1994, Reich et al. 2004, Reynolds et al. 2004, 
Wiens 2004). Because these objectives required a 
census of breeding goshawks, we attempted to find 
all breeding goshawks in our study area. In this 
paper, we first describe the sampling protocols we 
used to initially locate and monitor breeding gos- 
hawks on the Kaibab Plateau. We then describe the 


abundance and dispersion of breeding territories, 
the dispersion of alternate nests within territories, 
reproductive rates, and behaviors effecting gos- 
hawk detectability that resulted from 12 yr of im- 
plementing our protocols. Finally, we present boot- 
strap subsampling of our full samples of territories 
to estimate the number of breeding territories 
needed for precise estimates of the reproductive 
status and production of young by goshawks. Our 
purpose is to provide a framework for developing 
sampling protocols, sampling efforts, and sample 
sizes for demographic and habitat studies of gos- 
hawks in other populations. 

Study Area 

The study area (1728 km^) was all of the E^ibab Pla- 
teau above 2182 m elevation above sea level, and con- 
tained ca. 122 400 ha of ponderosa pine {Pinus ponderosa) 
forests between 2075-2450 m elevation, ca. 51 600 ha of 
mixed-conifer forests between 2450-2650 m elevation, 
and ca. 30 600 ha of spruce {Picea engelmannii) fir {Abies 
lasiocarpa) forests between 2650-2800 m elevation (Ras- 
mussen 1941, White and Vankat 1993). Pinyon {Pinus ed- 
ulis) juniper {Juniperus spp.) woodlands occurred below 
the study area between 1830-2075 m elevation and 
shrub-steppe occurred below 1830 m. With the exception 
of several narrow (<1 km) meadows, several areas 
burned by wildfire, and numerous tree harvest areas, for- 
ests on the study area were contiguous (Reynolds et al. 
1994, Joy et al. 2003). The southern one-third of the 
study area included the Grand Canyon National Park- 
North Rim (GCNP) and the northern two-thirds, the E.ai- 
bab National Forest (KNF) . Forests on the Kaibab Plateau 
are isolated from other forests by varying distances of 
shrub-steppe; the nearest forest to the north, 97 km; to 
the east, 250 km; to the west, 80 km; and to the south, 
89 km, with the exception of a small area of ponderosa 
pine forest on the south rim of the Grand Canyon at 18 
km (Reynolds et al. 2004). 

Methods 

Field Procedures. We defined a breeding territory as 
an area exclusively occupied by a pair of goshawks during 
a breeding season. This definition implied that territories 
were defended by resident goshawks, and the dispersion 
of breeding pairs was constrained by territoriality. While 
uncertain if or how territories were defended by gos- 
hawks, we estimated their size on the Kaibab Plateau as 
the area whose radius was half the mean distance among 
neighboring pairs. Recapture of marked goshawks 
showed that territorial owners, as well as their replace- 
ments over time, had strong life-time fidelity to their ter- 
ritory (Reynolds and Joy in press, R. Reynolds unpubl. 
data) , and territories on the Kaibab Plateau appeared to 
be spatially fixed over years. 


276 


Reynolds et al. 


VoL. 39, No. 3 


We located goshawk territories using two protocols: sys- 
tematic foot-searches for goshawks and their nests in ar- 
eas <1600 ha and broadcasts of goshawk vocalizations 
from stations on transects (Kennedy and Stahlecker 
1993, Joy et al. 1994) in areas >2400 ha. Both nest- 
searching procedures were used each breeding season 
(April-August). A new territory was identified when a 
used goshawk nest (or, in rare cases, an occupied-only 
nest area; see below) was discovered in an area not al- 
ready in a known territory and when the new nest (or 
nest area) was used by unbanded goshawks. Once a ter- 
ritory was found, it was added to that year’s cohort of 
territories and assessed in all subsequent years for gos- 
hawk occupancy. Because we were unable to search our 
study area completely in a single year, we extended our 
nest searching into previously unsearched areas each 
year; hence, the number of territories under study in- 
creased over years. In addition to expanded nest search- 
es, we annually re-searched areas (using both foot and 
broadcast searches) suspected of having territorial gos- 
hawks based on goshawk nest spacing (Reynolds and 
Wight 1978, Reynolds et al. 1994). 

Goshawk territories often contain one or more alter- 
nate nests that are used by the goshawks over several 
years (Squires and Reynolds 1997, Reynolds and Joy in 
press). To prevent misclassifying the reproductive status 
of goshawks that may have moved to an alternate nest, 
we used a within-territory nest-searching protocol con- 
sisting of three sequential steps (Reynolds et al. 2004). 
Each year, beginning 3 wk before egg laying, we con- 
ducted “initial visits” to all known alternate nests and 
historical nest structures (existing nests with unknown 
histories of use) to determine if goshawks were present. 
Searches for goshawks, their feces, molted feathers, and 
nests refurbished with green twigs (Reynolds and Wight 
1982) were conducted within 100-m radii of each alter- 
nate and historical nest. Initial visits to nests were com- 
pleted in all territories by 2 wk after egg-laying. If a used 
nest in a territory was not found in an initial visit, a “foot 
search” was conducted within a 500-m radius circle cen- 
tered on the last-used nest or the centroid of the territory 
(determined subsequent to discovery of >1 alternate 
nests in a territory). Territory centroids were the geo- 
metric means of coordinates of alternate nests weighted 
by the number of yr each alternate nest was used during 
our study (Reynolds et al. 2004, Reynolds and Joy in 
press) . A foot search involved systematically walking the 
500-m radius circle looking for goshawks or signs of their 
presence (see above). Foot searches were conducted 
from egg laying to about 15—20 d after egg-hatching. In 
territories where used nests were not located in foot 
searches, a “broadcast search” was conducted in a 1600- 
m radius circle centered on either the last-used nest or 
the territory centroid. Broadcasting of goshawk vocaliza- 
tions were conducted from stations on transects arranged 
as described by Joy et al. (1994). Broadcast searches were 
conducted from about 10 d after egg hatching to the end 
of the post-fledging dependency period (late August or 
early September). All nest trees were mapped to the 
nearest 3 m with a global positioning system. 

Nests were “used” if goshawks laid eggs, and territories 
were “occupied-only” if eggs were not laid but evidence 
(goshawks observed, molted feathers, feces, reconstruct- 


ed nest) of goshawk presence was found in association 
with a nest structure, or “unknown” if insufficient evi- 
dence of occupancy was found. All used nests were visited 
weekly to count numbers of nestlings and fledglings and 
to determine the approximate timing and causes of nest 
failure. Goshawk nestlings were banded in the 10 d be- 
fore fledgling, and numbers of nestlings present at the 
time of banding was considered the number of young 
produced. Nesting adults were captured with dho-gaza 
nets placed in nest areas and baited with live Great 
Horned Owls {Bubo virginianus) during the nestling pe- 
riod (Reynolds et al. 1994). All goshawks received a U.S 
Geological Survey leg band and a colored-aluminum leg 
band with a unique alpha-numeric code readable from 
80 ra with 40-60X telescopes (Reynolds et al. 1994). An- 
nual field efforts of crews consisting of 15-23 persons 
were focused on finding new territories, finding nests 
within known territories, and capturing and recapturing 
(or resighting) goshawks on the study area. 

Data Analysis. We used Dirichlet tessellation and De- 
launey triangulation (Cressie 1991) to estimate the dis- 
tances between the centroids of first-order neighboring 
goshawk territories. To estimate the dispersion of alter- 
nate nests within territories, we measured the within-ter- 
ritory map distances between each alternate nest (inter- 
alternate nest distance) and the within-territory centroid 
to each alternate nest (centroid-to-alternate nest dis- 
tance; excludes territories with only one nest). To test for 
differences in the spacing of goshawk territories in the 
KNF and the GCNP, we used a two-sample ^-test. To char- 
acterize the strength of the relationship between the 
numbers of new territories found in a year and the pro- 
portion of territories used in a year, we used a Spear- 
man’s correlation coefficient (rj. The annual proportion 
of territories with used nests was calculated as the pro- 
portion of those territories under study in the previous 
year (prior-year’s cohort of territories) that had used 
nests in the current year (Reynolds and Joy in press). We 
did this because the number of territories under study 
increased annually, and we included only territories that 
were monitored from before egg-laying to minimize bias 
associated with missed failed nests. We defined nest suc- 
cess as the proportion of used nests in a prior year’s co- 
hort of territories that produced >1 fledgling. To ex- 
amine annual differences between the proportion of 
territories with used nests and the mean number of 
young produced per used nest, we calculated 95% Con- 
fidence Intervals (Cl) for these parameters and visually 
assessed the degree of Cl overlap among estimates. 

We used the bootstrap method (Efron and Tibshiram 
1993) to estimate the number of goshawk territories that 
needed monitoring to attain precise estimates of the pro- 
portion of territories with used nests, nesting success, and 
number of young fledged per used nest. Our objective 
was to display variability in these parameters for different- 
sized samples given the full sample estimate. We con- 
ducted, with replacement, 1000 bootstrap iterations with 
sample sizes of 20, 40, 60, 80, 100, and 120 territories. 
We present the bootstrap results in box and whisker plots 
for only 2000 and 2002 because numbers of territories 
under study during those years were similar (120 and 
121), and 2000 was a relatively good breeding year (55% 
of territories had used nests) , while 2002 was a relatively 


September 2005 


Techniques 


277 


Table 1. Total territories, number of used nests (eggs laid), and number and percent of territories from previous 
year’s cohort of territories with used nests on the Kaibab Plateau, Arizona, 1991-2002. Previous-year’s territory cohorts 
were used because all territories in that cohort were monitored from before egg-laying in a current year, minimizing 
bias created by missing used nests due to early nest failure. 


Year 



1991 

1992 

1993 

1994 

1995 

1996 

1997 

1998 

1999 

2000 

2001 

2002 

Total territories 

37 

64 

82 

88 

99 

105 

106 

109 

113 

120 

120 

121 

Total used nests 
Used nests in 

36 

59 

67 

21 

53 

46 

31 

58 

57 

66 

30 

21 

previous year’s 
cohort 


32 

49 

18 

44 

40 

31 

55 

56 

60 

30 

21 

Percent used in 













cohort 


86 

77 

22 

50 

40 

30 

52 

51 

53 

25 

17 


poor breeding year (17% of territories had used nests). 
We plotted the medians of the estimates of the bootstrap 
subsampling for the proportions of territories with used 
nests and for nesting success. For mean young fledged 
per territory, we plotted the medians and coefficients of 
variation of the bootstrap estimates. We used a CV of 20% 
as a target criterion for identifying a level of sampling 
needed to attain adequate estimates of numbers of young 
per territory (see Pollock et al. 1990). 

Results 

The Study Population. Numbers of goshawk ter- 
ritories under study increased from 37 in 1991 to 
121 in 2002 as searches for new territories were 
annually extended into unsearched areas and as 
previously searched, but unoccupied areas, were 
searched again (Table 1). By 2002, about 95% of 
the KNF and 60% of the GCNP had been searched 
for nests. A total of 121 territories were discovered, 
and goshawks laid eggs in 1 or more years on all 
but six of these. Exceptions (two KNF, four GCNP 
territories) included territories occupied in ^2 yr 
by goshawks that built new, or reconstructed old, 
nests but did not lay eggs during the study. Terri- 
tory centroids were regularly spaced (Reich et al. 
2004, Reynolds and Joy in press) . The mean De- 
launay triangle distance between 120 territory cen- 
troids (1 territory not included due to inadequate 
search for surrounding territories) was 3.8 km (SD 
= 1.3 km, min = 1.3 km, max = 8.1 km, N— 302 
first-order neighbor distances; inter-centroid dis- 
tances that crossed unsearched areas in the ex- 
treme southeast of the study area were not includ- 
ed; Fig. 1). 

We estimated the total number of breeding ter- 
ritories on the study area by calculating an “exclu- 
sive” area for each pair of goshawks using one-half 
the mean distance between territory centroids (3.8 


km) as the radius and dividing the study area 
(173 200 ha) by the exclusive area (1134 ha; Reyn- 
olds and Joy in press). This should result in an 
accurate estimate of the total number of territories 
because of the regular spacing of territories 
(known for 80% of our study area) and because 
forests on the study area were nearly contiguous 
(Reynolds and Joy in press). The study area was 
large enough for there to be approximately 150 
territories, five territories more than our 1996 es- 
timate (Reynolds and Joy in press) . This increase 
reflected the discovery of 17 new territories be- 
tween 1997 and 2002 and a subsequent 0.1 km re- 
duction in the mean inter-centroid distance. 
Therefore, our sample of 121 known territories 
represented about 80% of the potential total num- 
ber of goshawk territories in our study area. 

Temporal and spatial variation in the frequency 
of egg-laying by goshawks on the study area was 
extensive. Temporal variation reflected periods of 
years with increasing or decreasing proportions of 
goshawks that laid eggs (Table 1, Fig. 2), and spa- 
tial variation reflected differences in the frequen- 
cies of egg-laying among territories (Table 2). In 
the 12 yr during which the 37 territories in the 
1991 cohort were monitored, 13 territories (36%) 
had used nests in <6 yr and 23 (64%) had used 
nests in >7 yr (1 territory never had a used nest), 
and of the 27 new territories found in 1992, 17 
(63%) had used nests in ^5 yr and 10 (37%) had 
eggs in >6 of the 11 yr they were monitored (Table 
2). Overall, 75% (86 of 115 territories that had 
used nests in ^1 yr) of territories had used nests 
in ^3 yr. Most (87%) territories in which egg lay- 
ing was skipped in ^1 yr had used nests or were 
occupied-only in subsequent years, often by the 


278 


Reynolds et al. 


VoL. 39, No. 3 


Grand 

Canyon 

National 

Park 


/V: : 

f • • A Jl.A 


Thiessen Polygons 
Delaunay Triangles 
Territory Centroids 






# i- • - . 

_ 1» ‘ 
mr '‘ 


> * « 


■w • 

a 


N 


Kaibab 

k '•. * • 

National 


Forest 
^ 

r ^ 















10 


20 



I km 


Figure 1. Thiessen polygons and Delaunay triangles used to calculate first-order nearest neighbor distances between 
Northern Goshawk territory centroids on the Kaibab Plateau, Arizona, 1991-2002 {N = 120, see text). Mean inter- 
centroid distance was 3.8 km (SD =1.3 km, min = 1.3 km, max = 8.2 km, N = 302 triangle legs). 


same banded goshawks that had previously laid 
eggs on the territory (R. Reynolds unpubl. data) . 

Of a combined total of 435 used nests in all 11 
prior-year cohorts of territories, 341 (63%) were 
successful (Table 3). Of 94 nest failures, 59 (63%) 
failed during the incubation period, and 35 (37%) 
failed during the nestling period. There was mini- 
mal among-year variation in nesting success (Table 
3, Fig. 2). In 459 broods (not limited to nests in 


prior-year cohorts) with accurate counts of young, 
brood sizes ranged from 1-4 nestlings (median = 
2; Table 2); 102 (22%) broods had one young, 219 
(48%) had two young, 133 (29%) had three young, 
and five (1%) had four young. The mean annual 
number of fledglings produced per used nest was 
only moderately variable compared to the annual 
variation in the proportion of territories with used 
nests; the CV of the number of young produced 




September 2005 


Techniques 




Year 


Figure 2. Annual variation in (a) the proportion of ter- 
ritories under study containing active (eggs laid) North- 
ern Goshawk nests and (b) in the mean numbers of 
young produced per used nest in the previous year’s co- 
hort of goshawk nests (see text) on the Kaibab Plateau, 
Arizona, 1991-2002. Error bars represent ±95% Cl. 


per used nest was 28%, while the CV of the pro- 
portion of territories with used nests was 114% 
(Fig. 2) . Likewise, the among-year variation in total 
young produced by the 1991 cohort of territories 
{N = 37) over 12 yr was also higher than the 
among-year variation of the means of young pro- 
duced per used nest for the same territories and 
years. Total young produced ranged from 16 in 
2002 to 65 in 1992 and had an among-year coeffi- 
cient of variation (CV) of 68%, and mean number 
of young produced per used nest ranged from 0.6 
in 2002 to 2.4 in 2000 and had a CV of 37%. Thus, 
both the annual proportion of territories with used 
nests and total young produced per year provide a 
more sensitive measure of the variable reproduc- 
tive output of goshawks than the annual mean 
number of young produced per used nest. 

Goshawk Behavior and Sources of Error. How 
well an estimate represents the true spatial distri- 
bution, density, or habitat occupancy of a species 


a 





1 









1 








1— H 

00 








i-H 










rH 

rH 
















o 

S' 

1 

1 







CM 

1 

1 
















CM 








cn 

CM 



1 






■' — ^ 


1 






00 

rH 





H 









CTl 












o> 

y«— V 



r V 




00 



1 

1 



o' 

5 




rH 



rH 


8 











s 

rH 







I— ( 

rH 

rH 

1 


1 







1 


1 





CO 

CO 




o 









u 









IZ) 









D 




^ ^ 









kO 


o 






CO 

rH 

CD 

m 


1 






^ ' 


1 



CM 


rH 

oo 



1 









0 







V 







OO 

00 




1 


rH 

oo 

CV 

1 



1 





1 





CV 

CV 

CM 

CO 


^ 


















o 









CQ 









§ 









P 




CV 

CO 



o 




1 

CV 

CV 

1 

1 

MO 





CD 

m 



OO 














in 

X V. 


00 




on 

t-H 

rH 

CD 

1 

rH 

1 



' — ^ 

'•' — ^ 

' 

1 

^ 

1 




CO 


rH 


CV 














,, — ^ 


rH 



s 



CM 

00 

rH 

rH 

1 


oo 



‘ — ■ 


' 

1 


' — ' 




CO 

in 

CV 


rH 

CM 













s 


CV 


0? 





1 

CV 


CV 

1— ( 





1 


'' — ^ 






CM 



i-H 

CO 

I-H 


c/i 


















2 

Q 








0 


.0 






w 

H 

p 

<o 

j> 

00 

to 

^H 

to 

2 

3 

0 

CO 

CV 

^H 


rH 



3 

Ph 

















H 

cd 

?— 1 

CV 

CO 


m 

to 




02 

Oi 


02 

Oi 

02 




02 


<Ji 

02 

O) 

02 




t-h 


rH 


rH 



V 

u 

biD 

g 

'C 

2 

3 

o 

6 


bo 

bo 


u 

<ri 

(-M 

o 

o 


(J 

tu 

Xi 

"0 

I 


o 

a 

CV 

o 

o 

CV 

O) 


u 

a; 

£ 

c 

0 

1 
& 

2 

a 

u 

u 

o 

Vi 

u 

2 

u 

<v 

o 


Table 3. Number of used (eggs laid) and successful (fledged >1 young) nests, mean and standard deviation (SD) of fledglings per used nest, and proportion 
of used nests within the prior year’s cohort of Northern Goshawk territories that fledged young on the Kaibab Plateau, Arizona, 1991-2002. 


280 


Reynolds et al. 


VoL. 39, No. 3 


(M 

O 

o 

CM 


o 

o 


o 

o 

o 

CM 


CJi 

CTS 


00 

C7i 


CJi 

as 


CO 

CJi 

as 


CJS 

os 


TP 

as 

CJS 


vn 

CJS 

os 


CM 

CJS 

CJS 


(M 


O 

CO 


o 

CO 


CO 


lO 

If) 


o 

00 


o 


tP 

tP 


C20 


O) 


CO 

00 00 
d d 


so 
Csf O 


os 


00 
J o 

c^i 1— I 00 

If) 


CM 




q 

CM 


CO 


CO 

tP 


so 
sf) oq 
i—i d so 
CM 


00 OS 
rH O 


oO 


CM 
00 Os 


O tP 
OO 


CM 
!-h Os 

I— I d CO 


Tp 

00 q 

1 — i ,-H CV 
tP 


(M 


CM 

SO 


OO 

CO 


00 

00 


00 


tP 

00 


00 

00 


00 




CM 


CO 

X 


CM 


CM 


X 


CO 




CM 






iS 






w 




4-1 


1) 

G 

(/) 

W 

cA 

t/3 

bo 

V 


3 

V 

1) 

c 




Cm 

1/3 

(/3 

w 

c 

Ti 

V 

V3 

1 — , 


<u 

c/3 

Q 

U 

3 

<u 

OJ 

p 



X 

X 

(1^ 


X 


- 5 


V 

u 

y 

:3 

c/3 


depends on the error associated with the estimate 
(Thompson et al. 1998). A potential source of sam- 
pling variation is an incomplete count of breeding 
goshawks. Counts are related to the actual size of 
the territorial goshawk population by the proba- 
bility of detection, which may vary systematically. 
Because of their defensive behavior at nests, the 
detectability of breeding goshawks (still relatively 
low due to their elusiveness and complex habitats) 
is much higher than the detectability of non-nest- 
ing goshawks (including those whose nests failed). 
Low detectability of nonbreeders combined with 
the large annual variation in numbers of goshawks 
breeding can produce large sampling variation. To 
account for the low detectability of nonbreeders, 
we repeatedly searched areas suspected of contain- 
ing goshawks. Eleven yr of repeated searching for 
nests showed that the KNF was nearly saturated 
with breeding territories (Fig. 1 ) . We do not know 
if the GCNP was similarly saturated with territories 
because only ca. 60% of the GCNP was searched 
for goshawks. However, the mean distance between 
centroids of known territories in the GCNP was not 
significantly different from the mean distance be- 
tween KNF centroids (KNF x = 3722 m, SD - 1221 
m, N — 271 triangle legs; GCNP x = 4028 m, SD 
-- 1477 m, N = 22 triangle legs; t — —1.1, df = 
291, P = 0.27), suggesting that the density of gos- 
hawk territories in the GCNP was similar to terri- 
tory density in the KNF. The success of finding new 
territories in a year was positively related to the 
proportion of territories with used nests in that 
year (r^ = 0.73, P = 0.011, N= 11); we found more 
new territories in good breeding years (1991-93, 
1998-2000) than in poor breeding years (1994, 
2002 ). 

Another source of measurement error is mis- 
classification of the breeding status of territorial 
goshawks. Detecting movements of goshawks 
among alternate nests required considerable sam- 
pling effort, the level of which depends on the 
number and distribution of alternates within ter- 
ritories and frequency of movement among the al- 
ternates. Because numbers of known alternates de- 
pends on years of monitoring, we only report the 
numbers of alternate nests in the 1991 and 1992 
cohorts of territories. Mean number of alternates 
in these territories was 3.2 (SD = 1.5 nests, min = 
1, max = 6, A = 36) and 2.9 (SD = 1.4 nests, min 
= 1, max = 6, A= 27), respectively. The frequency 
distribution of inter-alternate nest distances within 
all territories with >2 alternates {N = 91 territo- 


September 2005 


Techniques 


281 



Distance (m) 

Figure 3. Cumulative percent of alternate nests within 
territories with increasing distance (m) between territory 
centroids to alternate nests on the Kaibab Plateau, Ari- 
zona. Maximum centroid-to-alternate nest distance was 
1452 m. 

ries) was right-skewed with a median of 402 m (x 
= 612 m, SD = 569 m, min = 9 m, max = 2426 
m, = 308 alternate nests) . When measured from 
territory centroids, the median centroid-to-alter- 
nate distance was 228 m (x = 334 m, SD = 298 m, 
min = 6 m, max = 1452 m, = 273 alternates in 
91 territories), about half of the median inter-al- 
ternate nest distance. The cumulative proportion 
of alternates captured with distance from centroids 
showed that about 75% occurred within 0.5 km, 
and about 95% occurred within 1 km of centroids 
(Fig. 3). Thus, our territory-focused broadcast 


searching protocol in areas of 1.6-km radius 
around centroids exceeded the maximum known 
centroid-to-alternate distance (1452 m). The fre- 
quency of movement of egg-laying goshawks to al- 
ternates was high; an annual mean of 64% of 
breeding goshawks moved to an alternate, and 
42% of these movements were to new (unknown 
to us) alternates (Table 4). 

Sample Size. Bootstrap subsampling showed that 
samples of ca. 60—80 territories in good breeding 
years and 80-100 territories in poor breeding years 
were needed for precise estimates of the full sam- 
ple means of the proportion of territories with 
used nests and nesting success on the Kaibab Pla- 
teau (Fig. 4). Coefficient of Variation plots of the 
mean young per used nest in good breeding years 
showed that subsamples of >80 territories had 
bootstrap estimates entirely below a CV of 20%, 
although many of the estimates from subsamples 
of 60 territories were below 20% (Fig. 5) . In poor 
breeding years, subsamples of 100 territories were 
insufficient to achieve a CV of less than 20%, re- 
flecting the few (21) territories that were occcu- 
pied in 2002. How temporal and spatial variation 
in reproduction on the Kaibab Plateau compares 
to other goshawk populations is unknown because 
other studies typically reported reproduction at 
only used or successful nests (e.g., Reynolds and 
Wight 1978, DeStefano et al. 1994, Doyle and 
Smith 1994, Younk and Bechard 1994); only Keane 
et al. (in press) and Reynolds and Joy (in press) 


Table 4. Number (%) of breeding Northern Goshawks that stayed in the previous year’s nest or moved to a new 
or previously-used alternate nest within their territory on the Kaibab Plateau, Arizona, 1991-2002. 


Movement 

War 

Staved 

To New Alternate 

To Prior Alternate 

Total Percent 
Moving 

1992 

14 (45) 

17 (55) 

— 

55 

1993 

17 (35) 

26 (53) 

6 (12) 

65 

1994 

7 (39) 

7 (39) 

4 (22) 

61 

1995 

18 (43) 

17 (40) 

7 (17) 

57 

1996 

9 (24) 

16 (43) 

12 (32) 

76 

1997 

9 (30) 

14 (47) 

7 (23) 

70 

1998 

19 (35) 

27 (50) 

8 (15) 

65 

1999 

21 (38) 

18 (32) 

17 (30) 

63 

2000 

18 (30) 

20 (33) 

22 (37) 

70 

2001 

13 (43) 

10 (33) 

7 (23) 

57 

2002 

7 (33) 

8 (38) 

6 (29) 

67 

Total 

152 (36) 

180 (42) 

96 (22) 

64 


282 


Reynolds et al. 


VoL. 39, No. 3 


2000 


2002 a. 




b. 


.2 J3 

A V) 

O 0) 
C u 

S-i 

I* 



20 40 60 80 100 120 


20 40 60 80 100 120 


Number of Territories in Subsamples 


0 > 20 - 
£ 

3 

O 

>- 

-S 1-5- 


« 

.a 

E 


c 

n 

d) 


1 . 0 - 


0.5 


2000 


Be 


20 40 60 80 100 120 


0.5 - 
0.4 ■ 
0.3 
0.2 -\ 
0.1 
0.0 H 


2002 


bb- 


_ - ^ 4- 


20 40 60 80 100 120 


b 




Number Of Territories in Subsamples 


Figure 4. Box plots of bootstrap subsamples estimating 
the effects of sample size in good (2000) and poor 
(2002) breeding years on estimates of the proportion of 
Northern Goshawk territories with used nests (eggs laid) 
(a) and nesting success (b) on the Kaibab Plateau, Ari- 
zona, 1991-2002. Dotted vertical lines are numbers of 
territories (120 in 2000, 121 in 2002) used to estimate 
the true sample means (solid horizontal lines). Box plot 
whiskers extend to the maximum and minimum esti- 
mates, boxes represent the first and third quartiles of 
estimates, and the horizontal lines within boxes represent 
the medians of estimates. 


Figure 5. Box plots of bootstrap subsamples estimating 
the effects of sample size in good (2000) and poor 
(2002) breeding years on estimates of (a) the mean, and 
(b) the coefficient of variation (CV), of young produced 
per used (eggs laid) Northern Goshawk nest on the Kai- 
bab Plateau, Arizona, 1991-2002. Dotted vertical lines are 
numbers of territories (120 in 2000, 121 in 2002) used 
to estimate the true sample means (solid horizontal 
lines). Box plot whiskers extend to the maximum and 
minimum estimates, boxes represent the first and third 
quartiles of estimates, and the horizontal lines within 
boxes represent the medians of estimates. 


reported the extent of temporal variation in the 
proportion of pairs breeding. 

Discussion 

Goshatvk populations are difficult to enumerate 
and monitor because of their elusive behavior, rel- 
atively low densities, and their structur ally-complex 
forest habitats. Nonetheless, goshawk detectability 
increases during breeding (a 6-mo period) because 
of their aggressive nest defense. However, detect- 
ability of goshawks was highly variable among in- 
dividuals because of extensive temporal (year-to- 
year) and spatial (among territory) variation in 
breeding. Within a year, nonbreeding territorial in- 
dividuals have lower detectability than breeders, 
and among years, low-quality individuals (Wiens 
and Reynolds 2005) or individuals on low-quality 
territories have lower detectabilities than higher- 
quality individuals or those on higher-quality ter- 
ritories because they breed less often. Detectability 


within and among individuals can also be variable 
from year-to-year because of the number and dis- 
persion of alternate nests, and the frequency of 
movement among them. Finally, breeders whose 
nests fail have lower detectability than successful 
breeders. Therefore, determining the distribution, 
abundance, and habitat of a population of terri- 
torial goshawks and their annual breeding status 
requires sampling protocols and efforts that pro- 
vide for the detection of both breeding and non- 
breeding goshawks. Repeated nest searching of 
areas suspected of having breeding goshawks 
(“holes” based on territory spacing) eventually 
showed that our study area was saturated with 
breeding territories. Repeated searching was re- 
quired because as many as 8 yr elapsed on some 
territories between egg-laying. Not surprisingly, our 
success in locating territories depended on the 
quality of the breeding year; more new territories 


September 2005 


Techniques 


283 


were found in years when larger proportions of 
goshawks laid eggs. 

Nest searching did not cease with the discovery 
of a territory. Annually, between 50-75% of egg- 
laying goshawks moved to alternate nests within 
their territories, and in some years, more than half 
of these moves were to alternates unknown to us, 
some of which were more than 1 .4 km apart. Such 
movements have long been recognized as making 
the monitoring of breeding goshawks difficult 
(Woodbridge and Detrich 1994, Reynolds et al. 
1994, Kennedy 1997). In attempts to locate gos- 
hawks that may have changed nests, Kennedy 
(1997) and Woodbridge and Detrich (1994) 
searched 0. 7-1.0 km and 1.6 km around the pre- 
viously-used nest in a territory, respectively. If the 
distribution of alternates within territories on the 
Kaibab Plateau is representative of the distribution 
of alternates elsewhere, then these radii would con- 
tain 95 and 100% of alternate nests, respectively, 
but only if the nest last used was close to the center 
of the territory. However, the farther the last-used 
nest was from the center of a territory, the higher 
the probability of missing alternates with these ra- 
dii. This suggests that in the early years of a mon- 
itoring study, longer search radii should be used, 
at least until centroids of territories can be esti- 
mated. 

In studies of goshawk demography (e.g., Reyn- 
olds and Wight 1978, DeStefano et al. 1994, Reyn- 
olds et al. 1994, Kennedy 1997, Reynolds and Joy 
in press) and habitat (e.g., Bosakowski et al. 1999, 
Daw and DeStefano 2001, Finn et al. 2002, Joy 
2002, McGrath et al. 2003, La Sorte et al. 2004), 
valid inferences to the target population depend 
on an adequate temporal and spatial sampling. 
Our study showed that, because breeding is tem- 
porally and spatially variable and the detectability 
of nonbreeders is low, accurate estimates of the 
number and location of nests and territories de- 
pends on constancy in annual sampling efforts and 
numbers of years over which surveys are conduct- 
ed. Insufficient sampling for territories results in 
underestimates of breeding densities and habitat 
occupancy, and insufficient searches for nests with- 
in territories results in underestimates of annual 
proportions of pairs breeding and production of 
young. Because of large variation in the frequency 
of breeding, high rates of movement among nests, 
and low detectability of nonbreeders, it is particu- 
larly difficult to demonstrate unequivocally that 
territories are unoccupied in a year in which a used 


nest is not found. These factors, especially when 
combined with insufficient sampling, may result in 
an apparent decrease in territory occupancy and, 
ultimately, a population decline. The difficulty of 
confirming that a territory is actually unoccupied 
is the basis for our assigning territories with insuf- 
ficient evidence of occupancy as “unknown.” That 
territories continue to be occupied during non- 
breeding years was demonstrated by the fact that 
in many cases, the same color-marked goshawks 
were found to nest on the same territory before 
and after up to a 7 yr break in egg-laying (R. Reyn- 
olds unpubl. data) . Because of this, we suggest that 
“territory occupancy rate” (proportion of known 
territories occupied), a commonly used reproduc- 
tive parameter for goshawks (Crocker-Bedford 
1990, Kennedy 1997), may be a biased estimator of 
the number of breeders in a population. Finally, 
the frequency of movements among alternate nests 
suggests that the scale of measurement for deter- 
mining the breeding status and reproduction of 
goshawks should be at the territory level and not 
at the nest area. 

An objective of population monitoring is to ob- 
tain reliable estimates from samples to infer chang- 
es in a target population. Our bootstrapping re- 
sults showed that large samples of territorial 
goshawks (often larger than attained in many gos- 
hawk studies) were needed for precise estimates of 
the proportion of territorial goshawks breeding 
and their nesting success and reproduction. Large 
samples are needed because of the extensive an- 
nual variation in the proportion of territories with 
reproductive goshawks. Whether equally large sam- 
ples of territories or pairs of goshawks are needed 
for reliable estimates of these parameters in other 
populations will likely depend on whether these 
populations are as temporally and spatially variable 
in reproduction as the Kaibab Plateau population. 
DeStefano et al. (1994) in Oregon, Doyle and 
Smith (1994) in northwestern Canada, Wood- 
bridge and Detrich (1994) in northern California, 
Kennedy (1997) in New Mexico, and Keane et al. 
(in press) in central California, all reported mod- 
erate to extensive temporal variation in goshawk 
reproduction. Both the proportion of territories 
with egg-laying goshawks and total young produced 
on the Kaibab Plateau were more variable among 
years than mean numbers of young produced per 
used nest per year, the most commonly reported 
goshawk reproductive parameter (Kennedy 1997 
and references therein). Because the proportions 


284 


Reynolds et al. 


VoL. 39, No. 3 


of goshawks breeding and total young produced in 
a year more accurately portrayed the extent of an- 
nual variation in reproduction of the Kaibab gos- 
hawk population, both are likely to better describe 
a population’s response to fluctuations in resourc- 
es (e.g., food abundance; Salafsky 2004, Salafsky et 
al. 2005) and habitat quality than numbers of 
young produced per used nest. 

Conclusion 

Stratification of a study area, protocols for de- 
tecting species, and sampling efforts in studies are 
based on subjective and previous information 
(Morrison et al. 2001). Our nearly complete census 
of breeding goshawks on the Kaibab Plateau pro- 
vides information on the distribution, density, var- 
iation in reproduction, and breeding behavior of 
territorial goshawks in one population. Our intent 
in presenting these data was to provide a frame- 
work for developing sampling protocols and iden- 
tifying sampling efforts that may be needed to re- 
liably estimate the distribution, density, vital rates, 
and habitats of breeding goshawks in other popu- 
lations. Extensive temporal and spatial variation in 
reproduction on the Kaibab Plateau required as 
many as 8 yr of repeated nest searching to identify 
a population of breeders and annual searches of 
areas of 1 .4-km radius around territory centers for 
reliable estimates of the reproductive status of ter- 
ritorial pairs. Further, as many as 60—80 goshawk 
territories were needed for precise estimates of the 
annual production of young by a population. The 
specific sampling protocols and efforts used in our 
study, and the samples of territories identified in 
this paper, demonstrate that demography and hab- 
itat studies of goshawks may have to employ inten- 
sive and repeated searches for goshawks in large 
study areas over at least 8 yr. 

Acknowledgments 

This study was supported by the USDA Forest Service 
Southwestern Region, Kaibab National Forest, North Kai- 
bab Ranger District, and Rocky Mountain Research Sta- 
tion, and a Heritage Program grant from the Arizona 
Game and Fish Department. Many helped find, trap, and 
mark goshawks during this study. We especially thank J. 
Seyfried, C. Erickson, J. Lambert, D. Laing, M. Gavin, D. 
Leslie, R. Hadwin, S. Bayard de Volo, J. Burns, J. Fein- 
stein, A. Gillen, B. Hunt, L. Hunt, and C. Van Cleve for 
three or more years of help. We thank M. Bevers, C. 
Flather, R. King, and P. Lukacs and R. Steidl for helpful 
reviews of the manuscript. 


Literature Cited 

Bosakowski, T., B. McCullough, F.J. Lapsansky, and 
M.E. Vaugn. 1999. Northern Goshawks nesting on a 
private industrial forest in western Washington./. Rap- 
tor Res. 33:240-244. 

Cressie, N. 1991. Statistics for spatial data. John Wiley 
and Sons, Inc., New York, NYU.S.A. 

Crocker-Bedeord, D.C. 1990. Goshawk reproduction 
and forest management. Wildl. Soc. Bull. 18:262-269. 

Daw, S.K. and S. DeStefano. 2001. Forest characteristics 
of Northern Goshawk nest stands and post-fledging 
areas in Oregon. J. Wildl. Manag. 65:59-65. 

DeStefano, S., S.K. Daw, S.M. Desimone, and E.C. Mes- 
LOW. 1994. Density and productivity of Northern Gos- 
hawks: implication for monitoring and management. 
Stud. Avian Biol. 16:88—91. 

Doyle, F.I. and J.M.N. Smith. 1994. Population responses 
of Northern Goshawks to the 10-year cycle in num- 
bers of snowshoe hares. Stud. Avian Biol. 16:122-129. 

Efron, B. and RJ. Tibshirani. 1993. An introduction to 
the bootstrap. Monographs on statistics and applied 
probability 57. Chapman and Hall Press, London, 
U.K. 

Finn, S.P., J.M. Marzluff, and D.E. Varland. 2002. Ef- 
fects of landscape and local habitat attributes on 
Northern Goshawk site occupancy in western Wash- 
ington. Forest Sci. 48:427-436. 

Goodman, L.A. 1964. Simultaneous confidence intervals 
for contrasts among multinomial populations. Annals 
Math. Stat. 35:716-725. 

Joy, S.M. 2002. Northern Goshawk habitat on the Kaibab 
National Forest in Arizona: factors affecting nest lo- 
cations and territory quality. Ph.D. disseration, Colo- 
rado State Univ., Fort Collins, CO U.S.A. 

, R.M. Reich, and R.T. Reynolds. 2003. A non- 

parametric, supervised classification of forest types on 
the Kaibab National Forest using decision trees. Int 
J. Rem. Sens. 24:1835-1852. 

, R.T. Reynolds, and D.G. Leslie. 1994. Northern 

Goshawk broadcast surveys: hawk response variables 
and survey costs. Stud. Avian Biol. 16:24^30. 

Keane, J.J., M.L. Morrison, and M. Fry. In press. Tem- 
poral variation in Northern Goshawk reproduction in 
the Sierra Nevada, California. Stud. Avian Biol. 

Kennedy, P.L. 1997. The Northern Goshawk {Accipiter gen- 
tilis atricapillus) : is there evidence of a population de- 
cline? J. Raptor Res. 31:95-106. 

AND D.W. Stahlecker. 1993. Responsiveness of 

nesting Northern Goshawks to taped broadcasts of 
three conspecific calls./. Wildl. Manag. 57:249-257. 

La Sorte, F.A., R.W. Mannan, R.T. Reynolds, and TG. 
Grubb. 2004. Habitat associations of sympatric Red- 
tailed Hawks and Northern Goshawks on the Kaibab 
plateau./. Wildl. Manag. 68:307-317. 

McClaren, E.L., P.L. Kennedy, and S.R. Dewey. 2002. Do 


September 2005 


Techniques 


285 


some Northern Goshawk nest areas consistently 
fledge more young than others? Condor 104:343-352. 

McDonald, L.L. 2004, Sampling rare populations. Pages 
11-42 in W.L. Thompson [Ed.], Sampling rare or elu- 
sive species. Island Press, Washington, DC. U.S.A. 

McGrath, M.T., S. DeStephano, R.A. Riggs, L.L. Irwin, 
AND G.J. Roloff. 2003. Spatially explicit influences on 
Northern Goshawk nesting habitat in the interior Pa- 
cific Northwest. Wildl. Monogr. 154. 

Morrison, M.L., W.M. Block, M.D. Strickland, and 
W.L. Kendall. 2001. Wildlife study design. Springer, 
New York, NY U.S.A. 

Peterson, J.T. and P.B. Bayley. 2004. A Bayesian ap- 
proach to estimating presence when a species is un- 
detected. Pages 173-188 in W.L. Thompson [Ed.], 
Sampling rare or elusive species. Island Press, Wash- 
ington, DC U.S.A. 

Pollock, K.H., J.D. Nichols, C. Brownie, and J.E. 
Hines. 1990. Statistical inference for capture-recap- 
ture experiments. Wildl. Monogr. 104:1-97. 

Rasmussen, D.I. 1941. Biotic communities of Kaibab Pla- 
teau, Arizona. Ecol. Monogr. 11:229-275. 

Reich, R.M., S.M. Joy, and R.T. Reynolds. 2004. Predict- 
ing the location of Northern Goshawk nests: model- 
ing the spatial dependency between nest locations 
and forest structure. Ecol. Model. 176:109-133. 

Reynolds, R.T. and H.M. Wight. 1978. Distribution, den- 
sity, and productivity of accipiter hawks breeding in 
Oregon. Wilson Bull. 90:182—196. 

and . 1982. North American accipiter 

hawks. Pages 288-289 in D.E. Davis [Ed.], Handbook 
of census methods for terrestrial vertebrates. CRC 
Press, Boca Raton, FL U.S.A. 

, S.M. Joy, and D.G. Leslie. 1994. Nest productiv- 
ity, fidelity, and spacing of Northern Goshawks in Ar- 
izona. Stud. Avian Biol. 16:106—113. 

, G.C. White, S.M. Joy, and R.W. Mannan. 2004. 

Effects of radiotransmitters on Northern Goshawks: 


do tailmounts lower snrvival of breeding males? J. 
Wildl. Manag. 68:25-32. 

and S.M. Joy. In press. Demography of Northern 

Goshawks in northern Arizona, 1991-1996. Stud. Avi- 
an Biol. 

Salafsky, S.R. 2004. Covariation between prey abundance 
and Northern Goshawk reproduction on the Kaibab 
plateau, Arizona. M.S. thesis, Colorado State Univ., 
Fort Collins, CO U.S.A. 

, R.T. Reynolds, and B.R. Noon. 2005. Patterns of 

temporal variation in goshawk reproduction and prey 
resources. J. Raptor Res. 39:237-246. 

Squires, J.R. and R.T. Reynolds. 1997. Northern Gos- 
hawk {Accipiter gentilis). Pages 1—32 in A. Poole, and 
F. Gill [Eds.], The birds of North America. The Birds 
of North America, Inc., Philadelphia, PA U.S.A. 

Thompson, W.L., G.C. White, and C. Gowan. 1998. Mon- 
itoring vertebrate populations. Academic Press, Inc., 
San Diego, CA U.S.A. 

White, M.A. and J.L. Vankat. 1993. Middle and high el- 
evation coniferous forest communities of the north 
rim region of Grand Canyon National Park, Arizona, 
USA. Vegetatio 109:161-174. 

Wiens, J.D. 2004. Post-fledging survival and natal dis- 
persal of Northern Goshawks in Arizona. M.S. thesis, 
Colorado State Univ., Fort Collins, CO U.S.A. 

AND R.T. Reynolds. 2005. Is fledging success a re- 
liable index of fitness in Northern Goshawks 1 J. Raptor 
Res. 39:210-221. 

Woodbridge, B. and PJ- Detrich. 1994. Territory occu- 
pancy and habitat patch size of Northern Goshawks 
in the southern Cascades of California. Stud. Avian 
Biol. 16:83-87. 

Younk, J.V. AND MJ. Bechard. 1994. Breeding ecology of 
the Northern Goshawk in high-elevation aspen forests 
of northern Nevada. Stud. Avian Biol. 16:119-121. 

Received 26 February 2004; accepted 13 June 2005 

Guest Editor: Patricia L. Kennedy 

Associate Editor: Clint Boal 


J Raptor Res, 39 {S):2S6~295 
© 2005 The Raptor Research Foundation, Inc. 


POPULATION GENETICS AND GENOTYPING FOR MARK- 
RECAPTURE STUDIES OF NORTHERN GOSHAWKS (ACCIPITER 
GENTILIS) ON THE KAIBAB PLATEAU, ARIZONA 

Shelley Bayard de Volqi 

Graduate Degree Program in Ecology, Department of Biology, Colorado State University, Fort Collins, CO 80523 US. A. 
and Rocky Mountain Research Station, USDA Forest Service 2150 Centre Avenue, Building A, Suite 350, 

Fort Collins, CO 80526 US. A. 

Richard T. Reynolds 

Rocky Mountain Research Station, USDA Forest Service, NRRC 2150 Centre Avenue, Building A, Suite 350, 

Fart Collins, CO 80526 US. A. 

J. Rick Topinka 

Genomic Variation Laboratory, Department of Animal Science, Meyer Hall, University of California, Davis, 

One Shields Avenue, Davis CA 95616 US. A. 

Bernie May 

Genomic Variation Laboratory, Department of Animal Science, Meyer Hall, University of California, Davis, 

One Shields Avenue, Davis CA 95616 US. A. 

Michael F. Antolin 

Department of Biology, Colorado State University, Fort Collins, CO 80523 US. A. 

Abstract. — Advances in molecular techniques have facilitated use of genetic data in demographic wild- 
life studies. An important first step in genetic mark-recapture is selecting markers that uniquely “mark” 
and reliably “recapture” individuals. Markers should be tested on reliable DNAfrom known individuals 
(blood) before being used on non-invasively sampled DNA (hair, scat, or molted feathers) . To evaluate 
whether Northern Goshawks {Accipiter gentilis) can be uniquely identified by geno typing, 113 known 
(banded, sexed) goshawks from the Kaibab Plateau, Arizona, were genotyped using DNA from blood 
and five microsatellite markers and a sex-linked gene. We used mean relatedness to test whether adults 
in the population were related and probability of identity (P(id) = probability that two random individ- 
uals from the population have the same genotype) to test the ability of multi-locus genotyping for 
uniquely identifying goshawks. We used genetic data to assess inbreeding and demographic data to 
estimate the effective population size. Sixty-nine adult goshawks were sexed correctly and genotyped. 
Expected heterozygosity was high (H^ = 0.81), and relatedness among adults was low (r = —0.017). All 
individuals sampled (69 adults, 44 nestlings) had unique five-locus genotypes, the overall probability of 
identity was low (P(id) unbiased ~ ^7.03 X 10“^), and the observed P^jd) was <0.0001. Thus, Kaibab goshawks 
were uniquely “marked” by genotyping. Despite a small effective population size = 37 individuals), 
goshawks on the Kaibab Plateau functioned as a large breeding population with no inbreeding (Fjs = 
—0.001). We hypothesized that genetic diversity is maintained by gene flow via immigration of individuals 
from distant forests. 

Key Words: Northern Goshawk, Accipiter gentilis; capture-recapture, genetic tagging, individual identification', 

molecular sexing, probability of identity. 


genetica poblacional y genotipificacion para estudios de marcado-recaptura 

DE ACCIPITER GENTIUS EN KAIBAB PLATEAU, ARIZONA 

Resumen. — Los avances en las tecnicas moleculares han facilitado el uso de informacion genetica en 
estudios demograficos de fauna silvestre. Un primer paso importante en estudios geneticos de marcado 


* Corresponding author’s email address: sbdv@cnr.colostate.edu 


286 


September 2005 


Techniques 


287 


y recaptura es seleccionar marcadores que “marquen” inequivocamente y que permitan “recapturar” 
confiablemente a los individuos. Los marcadores deben ser probados en ADN confiable de individuos 
conocidos (sangre) antes de ser usados en muestreos no invasivos de ADN (pelo, excremento, o plumas 
mudadas) . Para evaluar si los individuos de la especie {Accipiter gentilis) pueden ser identificados por 
genotipificacion, estudiamos 113 gavilanes conocidos (sexados y anillados) de la meseta ELaibab, Arizona, 
usando ADN de la sangre y cinco marcadores microsatelites y un gen ligado al sexo. Usamos el par- 
entesco promedio para probar si los adultos en la poblacion estaban relacionados y la probabilidad de 
identidad (P(id) = la probabilidad que dos individuos al azar de la poblacion tengan el mismo genotipo) 
para probar la habilidad de la genotipificacion mediante multiples loci para identificar individualmente 
a los gavilanes. Utilizamos datos geneticos para determinar el grado de endogamia e informacion de- 
mografica para estimar el tamafio efectivo de la poblacion. Sesenta y nueve gavilanes adultos fueron 
correctamente sexados y genotipificados. La heterocigocidad esperada fue alta (H^ = 0.81) y el par- 
entesco entre adultos fue bajo (r = —0.017). Todos los individuos analizados (69 adultos, 44 polluelos) 
tuvieron genotipos unices con cinco loci, la probabilidad de identidad total fue baja (P(id) unbiased ~ 
7.03 X 10“^), y la P(id) observada fue <0.0001. Asi, los gavilanes de Kaibab fueron “marcados” singu- 
larmente por genotipificacion. A pesar de un tamano efectivo de la poblacion pequeno {N^ = 37 
individuos) , los gavilanes en el Kaibab Plateau funcionaron como una poblacion reproductiva grande y 
sin endogamia (Fjs = —0.001). Nuestra hipotesis es que la diversidad genetica es mantenida por flujo 
genetico a traves de la inmigracion de individuos de los bosques lejanos. 

[Traduccion de Mauricio Cotera] 


Northern Goshawks {Accipiter gentilis) are highly 
secretive and are most easily detected during the 
breeding season when they aggressively defend 
their nests and young. While their defensive be- 
havior at nests facilitates capture-recapture studies 
of breeding individuals, population monitoring is 
difficult because individuals often forgo breeding, 
or their nests fail early in a breeding season. Even 
in years of high productivity, mark-recapture stud- 
ies can be prohibitively expensive because popu- 
lation sampling requires large field crews and mul- 
tiple nest visits to many breeding territories to 
capture and recapture breeding goshawks (Reyn- 
olds et al. 2005). 

Because of recent improvements in molecular 
techniques (Haig 1998, Parkeuet al. 1998), genetic 
capture-recapture may be a viable alternative to tra- 
ditional capture-recapture methods for goshawks. 
Collecting molted feathers requires fewer nest vis- 
its than traditional capture-recapture methods. 
Breeding goshawks begin an annual molt during 
spring (Squires and Reynolds 1997) and because 
they spend much of the breeding season near their 
nests, they drop many of their molts within their 
nest areas, including years when nesting attempts 
fail. Thus, goshawk feathers are readily collected 
from nest areas and may provide an efficient 
means to non-invasively sample their populations. 

Several factors influence the success of genetic 
capture-recapture studies. An appropriate number 
of highly variable genetic markers for identifying 
individuals are required, and potential biases must 


be identified. “Shadow effects” (lack of discrimi- 
nation of individuals because of low variability or 
sampling too few markers) can negatively bias es- 
timates of population abundance and positively 
bias estimates of survival (Mills et al. 2000) . On the 
other hand, when more markers than necessary 
are used, population abundance may be overesti- 
mated and survival underestimated if genotyping 
errors add unique “genotypes,” and thus individ- 
uals, to population samples (Lukacs and Burnham 
2005). Both biases will inflate variance and lower 
precision of parameter estimates (Lukacs and 
Burnham 2005). 

Microsatellites are currently a preferred molec- 
ular marker for identifying individuals because 
they are easily interpreted (i.e., heterozygous ge- 
notypes are easily distinguished from homozygous 
genotypes) , highly variable, bi-parentally inherited, 
and generally appear to be selectively neutral. Fur- 
ther, a large body of literature exists on microsat- 
ellite evolution (Jarne and Lagoda 1996, Goldstein 
and Pollock 1997, Estoup et al. 2002), which has 
facilitated the development of much statistical the- 
ory and analytical software (Hedrick 2005). How- 
ever, microsatellites are expensive and time con- 
suming to develop for each newly-studied species. 
Occasionally primers used to amplify microsatellite 
markers in one species can be used in related spe- 
cies (Ellegren 1992, Primmer etal. 1996, Galbusera 
et al. 2000, Martinez-Cruz et al. 2002) . 

Prior to starting a non-invasive genetic study, es- 
tablishing intrapopulation genetic structure (i.e., 


288 


Bayard de Volo et al. 


VoL. 39, No. 3 


levels of inbreeding and relatedness) and the fre- 
quency of null alleles (alleles that fail to amplify) 
is necessary for providing baselines against which 
feather samples can be compared (Mills et al. 
2000). Likewise, it is important to establish statis- 
tical power of multi-locus genotyping for identify- 
ing individuals with an independent population 
sample. We present results from a pilot study where 
we assessed the feasibility of implementing a non- 
invasive genetic capture-recapture study on a pop- 
ulation of Northern Goshawks on the Raibab Pla- 
teau, Arizona. Before assessing the utility of molted 
feathers as a viable source of DNA, we established 
a dependable genotyping marker set using DNA 
derived from blood (Taberlet and Luikart 1999). 

Our objectives were to: (1) screen species-specif- 
ic and cross-specific (among species) microsatellite 
markers, (2) test a sex-linked gene in goshawks 
known to distinguish males and females in other 
raptors (Kahn et al. 1998), (3) assess the power of 
multi-locus genotyping to uniquely identify individ- 
uals using probability of identity analysis (P(id) 5 
probability that two individuals drawn at random 
from the same population share the same multi- 
locus genotype), and (4) estimate average relat- 
edness, inbreeding, and effective population size 
for the goshawk population on the Kaibab Plateau. 

Methods 

Field Collection. The goshawk study population is lo- 
cated on the Kaibab Plateau in northern Arizona, an area 
that includes the North Kaibab Ranger District of the 
Kaibab National Forest and the North Rim of the Grand 
Canyon National Park (for descriptions of the study area 
see Reich et al. 2004, Reynolds and Joy 2005). It is a 
forested plateau surrounded by shrub-steppe habitat — 
the nearest forests being 97 km to the north, 250 km to 
the east, 80 km to the west, and 89 km to the south — 
except for a small patch of forest 18 km south on the 
south rim of the Grand Canyon. Sampled nests were well 
distributed across the study area. We captured 69 adult, 
breeding goshawks (1991-93, 2000-02) and 44 of their 
nestlings (Reynolds et al. 1994). We sexed adult goshawks 
using morphometries (mass, tarsus length) and behavior. 
Blood was sampled from the brachial vein with 22-gauge 
needles and non-heparinized capillary tubes (volume ^ 
0.10 ml). Blood was transferred into STE (Sodium Chlo- 
ride-Tris-EDTA) buffer-filled storage tubes kept cool in 
insulated containers with frozen cold-packs until crews 
returned to the field station, where samples were subse- 
quently frozen (— 20°C). At the close of the field season, 
blood was transferred to and stored at — 80°C at Colorado 
State University, Fort Collins, CO. 

Laboratory Methods. We extracted DNA using 
QIAamp mini blood kits (Qiagen, Inc., Valencia, CA 
U.S.A.) following the manufacture’s protocol. To find mi- 
crosatellites, we screened published and unpublished 


primer sets that included microsatellites originally isolat- 
ed from Northern Goshawks (Topinka and May 2004), 
European goshawks {Aedpiter gentilis gentilis) , Golden Ea- 
gles {Aquila chrysaetos), and Red Kites {Milvus milvus, 
Peck 2000) . We also tested primers that amplify an intron 
within the avian CHD (chromo-helicase-DNA binding), 
which was used to determine gender in Red-tailed Hawks 
(Buteo jamaicensis) and Great Horned Owls {Bubo virgini- 
anus; Kahn et al. 1998). The CHD gene is located on the 
Z and W sex chromosomes. We expected males to be 
homozygous (ZZ genotype) and females to be heterozy- 
gous (ZW genotype). 

We used PCR (Polymerase-Chain-Reaction) to amplify 
microsatellites in 25 |xl reactions using 0.5 |xl (AGE la) 
or 1.0 [xl (all other markers) of template DNA, 2.5 pi 
10 X buffer containing 15 mM MggCL (Promega Corp. 
Madison, WI U.S.A.; for markers AGE la, AGE 2 and 
AGE 4 an additional 3 mM Mg 2 Cl 3 was added), 20 mM 
dNTPs, 25 pM each primer, lU Taq polymerase, and one 
drop mineral oil to prevent evaporation. Negative con- 
trols (reactions that include all reagents except template 
DNA) were included in every set of reactions, and we 
used “cold start” PCR where tubes (in racks) were kept 
on ice to prevent premature non-specific priming. We 
used MJR PTC-100 thermocyclers programmed for the 
following protocol: denature at 94°C for 4 min, 31 cycles 
of denature at 94°C for 40 sec, annealing at 58°C for 40 
sec, and chain extension at 72°C for 40 sec, with a final 
extension at 72°C for 5 min. 

We used PCR to amplify the CHD sex-linked gene in 
25pl reactions using l.Opl template DNA and the same 
reaction buffer described above. The PCR protocol in- 
cluded an initial 5 min at 95°C denature, 11 cycles of 
denature at 94°C for 30 sec, annealing at 52°C for 35 sec, 
and chain extension at 72°C for 2.0 min, 31 cycles of 
denature at 92°C for 30 sec, annealing at 56°C for 35 sec, 
and chain extension at 72°C for 2.0 min, with a final 
extension at 72°C for 7 min. 

We used gel electrophoresis to separate alleles. For mi- 
crosatellites, we used 8% polyacrylamide (Long Ranger, 
Cambrex Corp., Rockland, MA U.S.A.) denaturing gels 
(55 cm long) that were run at 45 watts for 4-5 hr, de- 
pending on allele size. For the CHD sex-linked gene, we 
used single-strand-conformation-polymorphism (SSCP) 
methods (Hiss et al. 1994) and electrophoresed alleles 
on non-denaturing gels at 5 watts for 15 hr. 

For microsatellites, we established allele standards us- 
ing representative samples from our first gel and then 
standardized all other gels using those same samples. 
Gels were scored visually, and allele standards were run 
on both sides of a gel to account for gel ambiguities that 
cause slight variations in migration distances. Further, a 
subset of individuals {N = 23) was genotyped a second 
time to validate scores for microsatellite markers. For the 
CHD marker, we ran all known females together {N = 
40) and all known males {N = 29) together to familiarize 
ourselves with allele morphology. Although not se- 
quenced, the fragments were ca. 240-260 base pairs, and 
Z and W alleles were similar in size, but were differenti- 
ated by the SSCP analysis (Hiss et al. 1994). 

Population Genetic Analysis. Population substructure, 
inbreeding, and genetic drift can reduce heterozygosity 
in populations. However, low yield and degraded DNA 


September 2005 


Techniques 


289 


sampled from sources such as molted feathers can arti- 
ficially reduce population heterozygosity if allelic drop- 
out (ADO; one of two alleles in a heterozygous individual 
fails to amplify) at one or more markers occurs. It is, 
therefore, important to use high yield sources of DNA 
(typically blood) from a known reference population to 
determine frequency of ADO (or null alleles) and true 
levels of heterozygosity (Taberlet et al. 1999). 

We used Cervus 2.0 (Marshall et al. 1998) to estimate 
observed (Hq) and expected (He) population heterozy- 
gosity and null allele frequencies. Cervus provides esti- 
mates of null allele frequencies with an iterative algo- 
rithm based on differences between observed and 
expected homozygote frequencies. We used Genepop 3.4 
(Raymond and Rousset 1995) to test for departures from 
Hardy-Weinberg equilibrium (random mating) and GDA 
1.0 (Lewis and Zaykin 2001) to test for linkage-disequi- 
librium (genotypes at one marker are independent from 
genotypes at other markers) and to estimate Fjg, an in- 
dicator of population substructure and inbreeding. For a 
review of F-statistics and microsatellite genetic markers, 
see Balloux and Lugon-Moulin (2002). 

To test our assumption that our sample of adult gos- 
hawks was not comprised of closely-related individuals, 
we used Identix 1.1 (Belkhir et al. 2002) to estimate 
mean pairwise relatedness. We used Queller and Good- 
night’s (1989) estimator option, and tested the null hy- 
pothesis of no relatedness by comparing our estimate to 
a distribution of coefficients derived through convention- 
al Monte Carlo resampling procedures (1000 permuta- 
tions) . 

Probability of Identity. The uniqueness of an individ- 
ual’s genotype depends on the number and polymor- 
phism (heterozygous) of the markers. Multi-locus geno- 
types based on few highly-variable markers can be as 
powerful as those based on many less variable markers 
(Waits et al. 2001). Mills et al. (2000) suggested for stud- 
ies of genetic demography that profiles should be based 
on multi-locus genotypes capable of discriminating indi- 
viduals with 99% certainty. Estimating probability of iden- 
tity (P(iD)) is one way to establish this certainty when it is 
expressed as 1 — P(id)' P(id) is similar to the match prob- 
ability used in human forensics (Evett and Weir 1998, 
Avise 2004, Hedrick 2005), but is less susceptible to vio- 
lations of linkage-disequilibrium and Hardy-Weinberg 
equilibrium, both of which can be prevalent in small, iso- 
lated, or substructured populations (Waits et al. 2001). 

P(iD) analysis includes two steps. First, two theoretical 
P(iD)’s, one for unrelated individuals (P(id) unbiased)) 
one for siblings (P(iD)sibs)> ^.re estimated (for equations 
see Waits et al. 2001). Both estimators use population 
allele frequency data, and P(iD)unbiased corrected for bias 
in small samples. The two estimators provide lower and 
upper confidence bounds on the number of markers 
needed to discriminate individuals accurately. If the study 
population is composed of many related individuals, then 
resolving those individuals requires more markers. Step 
two involves calculating an observed P(m)obs based on ac- 
tual multi-locus genotypes from a known population sam- 
ple and is simply the proportion of all possible pairs of 
individuals with identical multi-locus genotypes (Waits et 
al. 2001). 

To estimate both theoretical P(id)’s and to quantify 


P(iD)obs (the proportion of individuals that share geno- 
types), we used PROB-ID5 (Waits et al. 2001). We used 
multi-locus genotypes derived from 69 adult goshawks, 
which we assume to he unrelated (see below), and 44 of 
their nestlings (sibling groups of 2-4 nestlings) . We first 
analyzed the adults and then added the offspring/ sibling 
groups. We used all five microsatellite markers and the 
CHD sex-linked gene and added markers sequentially 
starting with those having the highest number of alleles. 

Effective Population Size. To evaluate whether immi- 
gration and gene flow influenced genetic structure of the 
Kaibab population we estimated its effective size (N^) 
Effective population size is the idealized number of in- 
dividuals in a population measured either demographi- 
cally, in terms how many individuals actually contribute 
to breeding (i.e., variance in productivity), or genetically, 
using F-statistics and measures of inbreeding where the 
assumption of non-overlapping generations exists (Bar- 
ton and Whitlock 1997, Hedrick 2005). Goshawks sam- 
pled on the Kaibab Plateau during the study period likely 
represent at least three overlapping generations; thus, we 
relied on reproductive data to estimate N^. 

Effective population size is generally smaller than the 
censused population {N). Counts of breeding pairs of 
goshawks can be used to index N^, but not all goshawk 
pairs on the Kaibab produced an equal number of off- 
spring during the study (Wiens and Reynolds 2005). We 
therefore estimated annual N^’s (equation 6.8a in Hed- 
rick 2005) as; 


where k is mean productivity measured as the number of 
young fledged per used nest, eggs laid, subsequent young 
fledged, or the nest failed (1991-2003; Reynolds et al. 
2005), 14 is the variance in annual mean productivity, and 
N is the annual count of breeding pairs for the year. We 
then calculated a 13-yr harmonic mean of annual A(.’s 
(equation 6.12b in Hedrick 2005) for our final size esti- 
mate. 

Results 

Genetic Markers. Of nine cross-specific and sub- 
specific markers tested, two did not resolve alleles 
successfully, six amplified successfully but lacked 
variability, and one both amplified and was poly- 
morphic (AGE la, Table 1). All four microsatellites 
originally isolated from Northern Goshawks ampli- 
fied and were polymorphic (Table 1). The CHD 
sex-linked gene amplified and SSCP genotypes 
were consistent within the sexes (females, N = 40; 
males N = 29) , making it useful for distinguishing 
between male and female goshawks. We validated 
our amplification and scoring of microsatellite 
markers after all individuals were genotyped and 
scored the first time. We genotyped the 23 individ- 
uals used as standards a second time using DNA 
that was archived and remained untouched in our 


Table 1. Statistics for microsatellites tested on blood-derived DNA sampled from female {N = 40) and male (N — 29) Northern Goshawks {Acapiter gentilis) , 
Kaibab Plateau, AZ (1991-1993, 2000-2002). 


290 


Bayard de Volo et al. 


VoL. 39, No. 3 


u 







o 

K 




1/2 

i-l 

U 


c/2 

W 

O 

M 

Cm 

cn 


CD eo 
CD 
eo O 

d d 


ID CD 1> O ^ 


ID 

ID 


(M 

o 9 . . . 

o o o o 
I I 




0 o 

1 I 




o 

o 


O 02 O ID ID 

O O O O O 

^ d d d d 

T— I 00 O CD CT5 CD 

CM O O p p 

d d d d d 


Tf 00 ID ID O 1—1 
OO 00 00 00 I> C30 

d d d d d d 


1> eo 00 D ^ 
1> 00 eo 00 J> Jt> 

d> <6 d> <6 d> d: 


ID CD ID CM CD 

CM O O CD O 

O 00 O ID 
P O 00 O O 

d d d d d 


iD 1“^ O iD Ocf 
T— 1 i-H rH iD 


M) lo M, 

s 

S3 52 « 

•Ci, -Cl, -2 

^ ^ ^ 

U Cj Cj Co .2 

P 'C 'S d s 

<«i*» 

cs c3 c e fciiO 


Csb bio bio fcuo 
roj 'tJ -ecj 


■o 

IM CD 


c/2 

<u 

<ri 


w w w w w 
O O O O O = 
< <! <3 ^ 


0. 

O 

CM 

Z 

O 


B 

O 

■D 

ce 

Mh 

<u 


o 

o 

o 


T5 

B 

rt 

cn 

<U 

D 

u 

MJ 

cO 

D 

O 

o 

CM 


B 

O 

N 

'u 

0 

B 

o 

S 

(L) 

T3 

O 

O 

O 


M3 

T3 

0 

D 


E 

■a 

D 

? 

O 

M 

u 

rt 


be 

E 


T3 

V 

’ii 

u 

T3 

CO 


I 

u 

X 

M IjJ 
cn 


PS 

u 


E 
0 
‘D 

a 

■E ^ 

B £f 

tci CJ 

II 

«5 4; 

4j ^ 
ert ^ 

'^3 ^ 

- E 

1 S 

E o 

C*H (h 

O ^ 

"c 

Sr B 

E P 

. 

Q ^ 

M M5 K2 


i 

C 


u 

a 

V 

u 

o 

o 

o 

o 


u 

<u 

M 


% 

bC 

c 

a 

Oh 

u 

4-1 

i2 

O 

O 

Xi 

be 

a 


T3 

OP 


IS 

u 


freezers from the time of original DNA extraction. 
We found only a single scoring error out of 230 
opportunities (23 samples genotyped twice for five 
markers). This was a recording error; the sample 
had actually genotyped correctly. 

In comparisons of expected (H^) and observed 
(Ho) heterozygosity at each marker, four of the five 
microsatellites were similar. However, one marker 
(AGE 6) significantly departed from Hardy-Wein- 
berg expectations (P < 0.01) due to a deficiency 
of heterozygote genotypes (Table 1). Based on our 
data, we suspected this marker was sex-linked, as 
we found strong linkage-disequilibrium between 
AGE 6 and the sex-linked CHD marker (P< 0.001, 
Fisher’s method, 3200 runs) indicating that the 
two markers segregate together. It appeared that 
the marker was located on the Z sex-chromosome 
because all females (ZW) had only a single allele 
(homozygous), while most males (ZZ) were hetero- 
zygous. We hypothesize that alleles on the female’s 
W-chromosome are non-amplifiable (null) because 
of mutations in the priming sequences flanking the 
marker (Scribner and Pearce 2000) or because the 
marker on the Z-chromosome simply has no ho- 
mologous region on the W-chromosome. We 
found no evidence for linkage-disequilibrium be- 
tween the other four microsatellite markers (Table 
1 ) when AGE 6 and CHD were excluded from the 
analysis. We found no evidence of null alleles, 
which is important for future assessments of ge- 
notyping error when using feathers as a source of 
DNA. 

Population Genetics. Fjs measures departures of 
observed and expected heterozygosity under as- 
sumptions of random mating and indicates either 
inbreeding (Fjs > zero) or inbreeding avoidance 
(Fis < zero). Thus, highly structured or isolated 
populations that experience genetic drift generally 
exhibit positive Fjs values. Alternatively, large pop- 
ulations or those experiencing high gene flow gen- 
erally exhibit nonsignificant or negative Fjs values. 

We found no evidence of inbreeding or inbreed- 
ing avoidance (Fjg = —0.001; 95% Cl = —0.070- 
0.063; AGE 6 excluded; Table 1), suggesting that 
Kaibab goshawks mate randomly. Lack of inbreed- 
ing could result from large population size, gene 
flow by immigrants, or both. However, our demo- 
graphically derived estimate of effective population 
size (N^ = 37 individuals; range = 10-86) indicated 
that the population was demographically small, 
thus making gene flow a more likely source of ge- 
netic variability. This is consistent with our estimate 


September 2005 


Techniques 


291 



Mean Coefficient of Relatedness 


Figure 1. Observed mean relatedness relative to 1000 
randomized populations assumed to lack relatedness. 
The observed mean falls below that which is expected at 
random, occurring with a probability of 3.7%, indicating 
that Northern Goshawks {Accipiter gentilis) on the Kaibab 
Plateau, Arizona, are less related than expected at ran- 
dom. 


of relatedness among adults goshawks, where mean 
relatedness (r^ = —0.017) was less than expected 
by random (Fig. 1). 

Probability of Identity. Certainty of individual 
identification is equal to 1 -P(id)j and therefore the 
goal in estimating probability of identity is to ob- 
tain small values of P(id)- Waits et al. (2001) sug- 
gested a value :^0.0001 for forensic investigations 
where estimates of demographic parameters are 
needed. This threshold is interpreted as a 1:10 000 
chance that two individuals sampled from the same 
population will have the same multi-locus geno- 
type. 

We found that all 69 unrelated adults had 
unique multi-locus genotypes with the inclusion of 
the first three markers (P(iD)obs 0.0001), and like- 
wise the estimated P(id) met the 0.0001 threshold 
(P(iD)unbiased = 1-13 X 10“^; Fig. 2a). With five 
markers, the same sample had a P(iD)unbiased ~ *7.03 
X 10“^. However, based on demographic data, we 
know that siblings and parent-offspring nested si- 
multaneously in the Kaibab population (R. Reyn- 
olds unpubl. data). To model this effect we added 
44 nestlings-siblings to the sample. While the two 
theoretical P(id)’s did not change, all five markers 
were required to differentiate individuals (P(iD)obs 
< 0.0001, Fig. 2b). 

In both cases (adult only and adults with off- 
spring-sibling groups) , our sample of markers was 
insufficiently large to bring the P(iD)sibs to the 
0.0001 threshold. Thus, we were not able to esti- 
mate an upper number of markers needed for this 


(a) 


g 


'+-( 

o 


(b) 


Q 




Number of Markers 

Figure 2. Relationship between theoretical, observed, 
and sib probability of identity (P(id)) for Northern Gos- 
hawks {Accipiter gentilis) on the Kaibab Plateau, Arizona. 
The first five markers are microsatellites and the sixth is 
the CHD sex-linked gene. Observed data closely tracked 
that of the theoretical estimator; however, while (a) all 
69 unrelated adult goshawks were resolved after the first 
three markers (observed P(jd) < 0.0001), (b) it took an 
additional two markers to resolve sibling and parent-off- 
spring goshawks when 44 nestlings were added to the 
sample. In both cases, the theoretical P(id) met our 
0.0001 threshold (a 1:10 000 chance that two individuals 
sampled from the same population have identical multi- 
locus genotypes). However, we did not analyze enough 
markers to bring the sibling P(id) to the threshold level. 


resolution. Nonetheless, with five markers P(iD)sibs 
= 6.17 X 10 which translates into a six in a 1000 
chance of drawing two identical genotypes. Be- 
cause we sampled many parent-offspring pairs that 
we could nevertheless distinguish, we are confident 
the five combined markers provide unique genetic 
marks. 

Discussion 

Our intent in this study was to develop a set of 
genetic markers that uniquely identified individual 
goshawks. We desired to establish this marker set 
using high-yield DNA (blood) sampled from a 


292 


Bayard de Volo ex al. 


Vox. 39, No. 3 


known reference population. While most microsat- 
ellites tested did not amplify or were monomor- 
phic (most A. g. gentilis markers) , we did find a set 
of highly variable markers that consistently ampli- 
fied DNA from blood. Elsewhere (S. Bayard de 
Volo unpubl. data) , we found that the same genetic 
markers consistently and reliably amplified DNA 
from molted feathers. We note that because AGE 
6 is probably on the Z sex chromosome, its utility 
for estimating within population relatedness and 
levels of inbreeding is limited to samples from 
males. However, differences in allele frequencies 
between populations will still be useful for larger- 
scale studies comparing populations. In a study of 
goshawks in Utah, Sonsthagen (2002) used a dif- 
ferent set of microsatellite markers than ours. Of 
eight markers, only one of theirs exhibited the 
same number of alleles (11 alleles, Hq = 0.73, Hjg 
= 0.74) as AGE 6 did in our study. This alternative 
marker (BV 20; Gautschi et al. 2000) would be use- 
ful if it exhibited similar levels of heterozygosity 
and allelic diversity in the Kaibab population. Re- 
placing AGE 6 with a less variable marker would 
result in having to add more markers to the entire 
genetic profile, which would introduce more op- 
portunities for genotyping error. We are currently 
testing BV 20 to see if it is an effective replacement 
for AGE 6. 

With the five microsatellite markers tested (Ta- 
ble 1), all 113 goshawks sampled had unique multi- 
locus genotypes resulting in a P(iD)obs 0.0001 
(Fig. 2b) and a P(iD)unbiased = 7.03 X 10-'^. This was 
a powerful result considering that our sample in- 
cluded many parent-offspring and sibling pairs 
from the same nest. Likewise, the five microsatel- 
lites showed a high level of expected heterozygosity 
(He = 0.81). Others have shown that marker sets 
composed of five markers that result in Hg s 0.80 
will have a theoretical P(id) ^ 0.0001 (Waits et al. 
2001). In Paetkau’s (2003) retrospective analysis of 
21 non-invasive genetic studies in bears {Ursus 
americanus, U. arctos), the number of markers used 
was determined by whether the first five most-var- 
iable microsatellite markers together had Hg > 
0.80. He found that for some black bear popula- 
tions He was >0.80 for five markers; however, for 
others, and for all grizzly bear populations, that Hg 
was <0.80, requiring the marker set to be in- 
creased to six or seven loci in order to discriminate 
among individuals. 

We note that our estimates of P(id) are specific 
to the Kaibab goshawk population; we cannot pre- 


dict with complete certainty that these same mark- 
ers will uniquely mark goshawks from other pop- 
ulations. Power of discrimination depends on 
population-specific levels of genetic variability (het- 
erozygosity) ; goshawk populations that are less var- 
iable because of geographic isolation or habitat 
fragmentation may require more markers to 
uniquely genotype individuals (Paetkau and Stro- 
beck 1994). However, goshawks are highly vagile, 
and we suspect gene flow is high among popula- 
tions. These goshawk populations will probably ex- 
hibit similar heterozygosity, and the marker set test- 
ed here should prove useful for other studies. 

The Kaibab goshawk population exhibits high 
genetic variability (Table 1 ) , despite its geographic 
isolation and small effective population size (based 
on demographic data; 13-yr x = 37 individuals). 
Several explanations may account for this. First, it 
is possible that the markers used in this study are 
under selective sweeps with genes that are affected 
by balancing selection for heterozygous genotypes. 
Such selection has been found for the genes of the 
MHC (major histocompatability complex) in mam- 
mals (Avise 2004), in which heterozygous individ- 
uals experience a fitness advantage. However, we 
suggest it is unlikely that all four nonsex-linked 
markers would be under the same selective pres- 
sures, given that they exhibit independent segre- 
gation (no evidence of linkage disequilibrium; see 
Black et al. 2001). 

A second and more likely explanation is that ac- 
tual for this goshawk population is much larger 
because geographically distant populations in the 
region are connected by migration and gene flow. 
While adult goshawks are mostly sedentary on 
breeding territories (Detrich and Woodbridge 
1994, Squires and Ruggiero 1995, Reynolds and Joy 
2005), band recoveries of first-year goshawks from 
the Kaibab Plateau indicate dispersal distances of 
up to 440 km (Wiens 2004) . In addition, telemetry 
data show that juvenile goshawks disperse from the 
Kaibab Plateau in their first year, with the majority 
moving beyond the 80 km detection distance 
(Wiens 2004). Further, Wiens (2004) showed that 
only 11% of 614 banded nestlings returned to be 
recruited into the Kaibab breeding population, in- 
dicating high first-year mortality or low natal-site 
fidelity. Evidence for the latter is indicated by the 
lack of population genetic structure for goshawks 
in Utah (Sonsthagen et al. 2004). To better assess 
actual effective population size for goshawks in 
western North America, we are expanding our 


September 2005 


Techniques 


293 


studies to include populations in the western por- 
tion of the species range. Data from these studies 
should allow a more comprehensive evaluation of 
the genetic structure and effective population size 
for goshawks in the West. 

Conclusions 

Genetic marking of Northern Goshawks on the 
Kaibab Plateau is both feasible and reliable. Like- 
wise, non-invasive genetic sampling will provide an 
alternative method for demographic and genetic 
data collection, as we have found that molted 
feathers are as reliable a source of DNA as blood 
(S. Bayard de Volo unpubl. data). Because gos- 
hawks show high territory fidelity (e.g., Detrich 
and Woodbridge 1994, Reynolds and Joy 2005), 
they are particularly well suited for non-invasive ge- 
netic sampling. We recommend that monitoring 
programs implement rigorous field collection of 
molted feathers. As with any demographic study, 
valid inferences to the population depend on ap- 
propriate spatial and temporal sampling from that 
population. Researchers and managers interested 
in implementing non-invasive genetic mark-recap- 
ture to study goshawks should contact the corre- 
sponding author or refer to Bayard de Volo (2005), 

Acknowledgments 

Laboratory research was funded through the Small 
Grant Research Fund, Grand Canyon National Park 
Foundation; Graduate Degree Program in Ecology Small 
Research Grant; Colorado State University Graduate Re- 
search Grant; and the 2002 Stephen R. Tully Memorial 
Grant, Raptor Research Foundation. Field sampling was 
funded by the USDA Forest Service Southwest Region 
and the Rocky Mountain Research Station. This work 
would have not been possible without the many field 
crew members who worked tirelessly to locate nests, and 
we especially thank M. Gavin, S. Joy, D. Laing, J. Seyfried, 
and D. Wiens for their undaunted efforts in trapping gos- 
hawks and collecting blood samples. L. Savage and D. 
Tripp provided invaluable expertise in laboratory proce- 
dures, and N. Stifani assisted with collecting genetic data. 
We thank R. King for statistical review and K. Burnham, 
M. Douglas, and R. Lopez, as well as two anonymous re- 
viewers for comments on early drafts of the manuscript. 

Literature Cited 

Avise, J.C. 2004. Molecular markers, natural history and 
evolution. Sinauer Associates, Inc., Sunderland, MA 
U.S.A. 

Balloux, F. and N. Lugon-Moulin. 2002. The estima- 
tion of population differentiation with microsatellite 
markers. Mol. Ecol. 11:155-165. 

Barton, N.H. and M.C. Whitlock. 1997. The evolution 
of metapopulations. Pages 183—209 in I. Hanski and 


M.E. Gilpin [Eds.], Metapopulation biology: ecology, 
genetics and evolution. Academic Press, San Diego, 
CA U.S.A. 

Bayard de Volo, S. 2005. Molt collection protocol, 
http:/ /lamar. colostate.edu/~sbdv/protocols_page/ 
protocols.htm. 

Belkhir, K., V. Gastric, and F. Bonhomme. 2002. Iden- 
Tix, a software to test for relatedness in a population 
using permutation methods. Mole. Ecol. Notes 2:611- 
614. 

Black, W.C., IV, C.F. Baer, M.E Antoun, and N.M. 
DuTeau. 2001. Population genomics: genome-wide 
sampling of insect populations. Ann. Rev. of Entomology 
46:441-469. 

Block, W.M., M.L. Morrison, and M.H. Reiser. 1994. 
Symposium overview: introduction. Stud. Avian Biol. 
16:1-2. 

Detrich, P. J. and B. Woodbridge. 1994. Territory fidel- 
ity, mate fidelity, and movements of color-marked 
Northern Goshawks in the southern Cascades of Cal- 
ifornia. Stud. Avian Biol. 16:130—132. 

Ellegren, H. 1992. Polymerase-chain-reaction (PCR) 
analysis of microsatellites — a new approach to studies 
of genetic relationships in birds. Auk 109:886—895. 

Estoup, a., P. Jarne, and J.-M. Cornuet. 2002. Homo- 
plasy and mutation model at microsatellite loci and 
their consequences for populations genetics analysis. 
Mole. Ecol. 11:1591-1604. 

Evett, I.W. and B.S. Weir. 1998. Interpreting DNA evi- 
dence: statistical genetics for forensic scientists. Sin- 
auer Associates, Inc. Sunderland, MA U.S.A. 

Galbusera, R, S. van Dongen, and E. Matthysen. 2000. 
Cross-species amplification of microsatellite primers 
in passerine birds. Conserv. Genetics 1:163—168. 

Gautschi, B., I. Tenzer, J.P. Muller, and B. Schmid. 
2000. Isolation and characterization of microsatellite 
loci in the bearded vulture {Gypaetus barbatus) and 
cross-amplification in three old world vultures. Mol. 
Ecol. 9:2193-2195. 

Goldstein, D.B. and D.D. Pollock. 1997. Launching mi- 
crosatellites: a review of mutation processes and meth- 
ods of phylogenetic inference. J! Heredity 88:335-342. 

Haig, S.M. 1998. Molecular contributions to conserva- 
tion. Ecol. 79:413-425. 

Hedrick, P.W. 2005. Genetics of populations. Jones and 
Bartlett Publishers, Inc., Sudbury, MA U.S.A. 

Hiss, R.H., D.E. Norris, C.H. Dietrich, R.F. Whitcomb, 
D.F. West, C.F. Bosto, S. Kambhampati, J. Piesman, 
M.F. Antolin, and W.C. Black, IV. 1994. Molecular 
taxonomy using single-strand conformation polymor- 
phism (SSCP) analysis of mitochondrial ribosomal 
genes. Insect Mol. Biol. 3:171—182- 

Jarne, P. and P.J.L. Lagoda. 1996. Microsatellites, from 
molecules to populations and back. Tree 11:424-429. 

Kahn, N.W., J. St. John, and T.W. Quinn. 1998. Chro- 
mosome-specific intron size differences in the avian 


294 


Bayard de Volo et al. 


VoL. 39, No. 3 


CHD gene provide an efficient method for sex iden- 
tification in birds. Auk 115:1074-1078. 

Lewis, P.O. and D. Zaykin. 2001. Genetic data analysis: 
computer program for the analysis of allelic data. Ver- 
sion 1.0 (dl6c). 

Lukacs, P.M. and K.P. Burnham. 2005. Estimating pop- 
ulation size from DNA-based capture-recapture data 
incorporating genotyping error. J. Wildi Manag. 69: 
396-403. 

Luikart, G. and P.R. England. 1999. Statistical analysis 
of microsatellite DNA data. Tree 14:253-256. 

Marshall, T.C., J. Slate, L.E.B. Kruuk, and J.M. Pem- 
berton. 1998. Statistical confidence for likelihood- 
based paternity inference in natural populations. Mol. 
Ecol. 7:639-655. 

Martinez-Cruz, B., V.A. David, J.A. Godoy, J.J. Negro, 
S.J. O’Brian, and W.E. Johnson. 2002. Eighteen poly- 
morphic microsatellite markers for the highly endan- 
gered Spanish Imperial Eagle (Aquila adalberti) and 
related species. Mol. Ecol. Notes 2:323-326. 

Mills, L.S., JJ- CiTTA, K.P. Lair, M.K. Schwartz, and 
D.A. Tallmon. 2000. Estimating animal abundance 
using non-invasive DNA sampling: promise and pit- 
falls. Ecol. Appl. 10:283-294. 

Nesje, M. and K.H. R0ED. 2000. Microsatellite DNA mark- 
ers from the Gyrfalcon (Falco rusticolus) and their use 
in other raptor species. Mol. Ecol. 9:1433-1449, 

Paetkau, D. 2003. An empirical exploration of data qual- 
ity in DNA-based population inventories. Mol. Ecol. 12: 
1375-1387. 

AND C. Strobeck. 1994. Microsatellite analysis of 

genetic variation in black bear populations. Mol. Ecol. 
3:489-495. 

Parker, P.G., A.A. Snow, M.D. Schug, G.G. Booton, and 
O.A. Euerst. 1998. What molecules can tell us about 
populations: choosing and using a molecular marker. 
Ecology 79:361-382. 

Peck, N. 2000. DNA forensics of raptors and the isolation 
and characterization of microsatellite markers in Ac- 
cipitridae. Ph.D. dissertation, Univ. of Nottingham, 
Nottingham, U.K. 

Primmer, C.R., A.P. Moller, and H. Ellegren. 1996. A 
wide-range survey of cross-species microsatellite am- 
plification in birds. Mol. Ecol. 5:365-378. 

Queller, D. and K. Goodnight. 1989. Estimating relat- 
edness using genetic markers. Evolution 43:258-275. 

Raymond, M. and E. Rousset. 1995. GENEPOP Version 
3. Id; population genetics software for exact tests and 
ecumenicism. y. Heredity 86:248-249. 

Reich, R.M., S.M. Joy, and R.T. Reynolds. 2004. Predict- 
ing the location of Northern Goshawk nests: model- 
ing the spatial dependency between nest locations 
and forest structure. Ecol. Modeling 176:109—133. 


Reynolds, R.T. and S.M. Joy. 2005. Demography of 
Northern Goshawks in northern Arizona, 1991-1996. 
Stud. Avian Biol. In press. 

, , AND D.G. Leslie. 1994. Nest productivity, 

fidelity, and spacing of Northern Goshawks in north- 
ern Arizona. Stud. Avian Biol. 16:106-113. 

, J.D. Wiens, S.M. Joy, and S.R. Salafsky. 2005. 

Sampling considerations for demographic and habitat 
studies of goshawks. J. Raptor Res. 39:274-285. 

Scribner, K.T. and J.M. Pearce. 2000. Microsatellites: 
evolutionary and methodological background and 
empirical applications at individual, population and 
phylogenetic levels. Pages 235—273 mA.J. Baker [Ed.], 
Molecular methods in ecology. Blackwell Science Ltd., 
Oxford, U.K. 

Sonsthagen, S.A. 2002. Gene flow and genetic charac- 
terization of Northern Goshawks {Accipiter gentilis) 
breeding in Utah, USA. M.S. thesis, Brigham Young 
Univ., Provo, UT U.S.A. 

, S.L. Talbot, and C.M. White. 2004. Gene flow 

and genetic characterization of Northern Goshawks 
{Accipiter gentilis) breeding in Utah. Condor 106:826— 
836. 

Squires, J.R. and R.T. Reynolds. 1997. Northern Gos- 
hawk {Acdpiter gentilis) . In A. Poole and F. Gill [Eds.], 
The birds of North America, No. 298. The Birds of 
North America, Inc., Philadelphia, PA U.S.A. 

AND L.F. Ruggiero. 1995. Winter movements of 

adult Northern Goshawks that nested in south-central 
Wyoming./. Raptor Res. 29:5-9. 

Taberlet, P. and G. Luikart. 1999. Non-invasive genetic 
sampling and individual identification. Biol. J. Linn. 
Soc. 68:41-55. 

, L.P. Waits, and G. Luikart. 1999. Noninvasive 

genetic sampling: look before you leap. TREE 14:323- 
327. 

Topinka, J.R. AND B. May. 2004. Development of poly- 
morphic loci in the Northern Goshawk (Accipiter 
gentilis) and cross-amplification in other raptor spe- 
cies. Conserv. Genetics 5:861—864. 

Waits, L.P., G. Luikart, and P. Taberlet. 2001. Estimat- 
ing the probability of identity among genotypes in 
natural populations: cautions and guidelines. Mol. 
Ecol. 10:249-256. 

Wiens, J.D. 2004. Post-fledging survival and natal dis- 
persal of juvenile Northern Goshawks in Arizona. M.S. 
thesis, Colorado State University, Fort Collins, CO 
U.S.A. 

AND R.T. Reynolds. 2005. Is fledging success a re- 
liable index of recruitment in Northern Goshawks? J. 
Raptor Res. 39:210-221. 

Received 26 February 2004; accepted 15 May 2005 

Guest Editor: John Keane 


September 2005 


Techniques 


295 


Appendix. Source, repeat qualities and accession numbers or primer sequences for microsatellites found to be useful 
for genotyping Northern Goshawks {Accipiter gentilis) on the Kaibab Plateau, Arizona. 


Marker 

Repeat Motif^ 

Allele Size 
IN Base 
Pairs 

Accession No. 

OR Primer Sequences 
5' TO 3' 

Author 

AGE 1 

(gggaalg. . (gaga)g. . (gagaala 

216 

AY312451 


AGE 2 

(gagaa) io(ga)4 

170 

AY312452 

Topinka and 

AGE 6 

( gagaa ) 4 . . ( gagaa ) 2 . • ( gagaa ) 5 

259 

AY312456 

May 2004 

AGE 4 

(gagaa) 

275 

AY312454 


AGE la 

(ggat) s 

208*^ 

f acaactgggctgtgctttgc 
r cttcccggtggctgaggctt 

Peck 2000 


® Sequenced by authors. 

Mean allele size in European goshawk (A. g. gentilis) . 


J. Raptor Res. 39(3):296-302 
© 2005 The Raptor Research Foundation, Inc. 

WHEN ARE GOSHAWKS NOT THERE? IS A SINGLE VISIT 
ENOUGH TO INFER ABSENCE AT OCCUPIED NEST AREAS? 

Douglas A. Boyce, W 

USDA Forest Service, Pacific Northwest Research Station, 2770 Sherwood Lane, Juneau, AK 99801 U.S.A. 

Patricia L. Kennedy 

Eastern Oregon Agricultural Research Center & Department of Fisheries and Wildlife, Oregon State University, 

P.O. BoxE, Union, OR 97883 US. A. 

Paul Beier 

School of Forestry, Northern Arizona University, Flagstaff, AZ 86011-5018 U.S.A. 

Micheal F. Ingraldi 

Arizona Game and Fish Department, Research Branch, 2221 W. Greenway Road, Phoenix, AZ 85023 U.S.A. 

Susie R. MacVean 

Arizona Game and Fish Department, 3500 South Lake Mary Road, Flagstaff, AZ 86004 U.S.A. 

Melissa S. Siders 

USDI Bureau of Land Management, Grand Staircase-Escalante National Monument, 190 East Center Street, 

Kanab, UT 84741 U.S.A. 

John R. Squires 

University of Montana Forestry Sciences Laboratory, 800 East Beckwith, Missoula, MT 59807 U.S.A. 

Brian Woodbridge 

USDI Eish and Wildlife Service, 1829 South Oregon Street, Yreka, CA 96097 U.S.A. 

Abstract. — ^We tested the efficacy of three methods (historical nest search, broadcast search, and tree 
transect search) for detecting presence of the Northern Goshawk (Accipiter gentilis) at occupied nest 
areas during the 1994 breeding season using only a single visit to a previously known nest area. We used 
detection rates in a probability model to determine how many visits are required to have confidence in 
reporting absence of goshawks. The purpose of this study is to understand if the three methods for 
detecting goshawks are robust enough for managers to rely on them for making land management 
decisions that may impact goshawk nest areas. Blind tests were conducted throughout the western 
United States. Results were similar among methods with goshawk presence going undetected at 36-42% 
of the occupied nest areas after a single visit. These results indicate that a single visit to a nest area is 
inadequate to provide reliable information on nest area occupation. Our probability of detection model 
showed that if each detection method is repeated three (historical or tree transect) or four (broadcast) 
times, goshawk absence can be inferred with a high level of confidence. Conclusions regarding nest 
area occupation using a single visit sampling method should be made with utmost caution. Classifying 
a nest area as vacant, when in fact goshawks are present, is a serious concern and leads to spurious 
conclusions. Land managers making habitat-altering decisions should not rely on a single visit to nest 
areas to establish the absence of goshawks. Possibilities for improving the detection of nesting goshawks 
include multiple independent visits using the same method, using a sequence of techniques in combi- 
nation to yield an improved cumulative probability of detection, or developing a new method yielding 
a higher probability of detection. The historical nest search obtained the best results, followed by the 
tree transect and broadcast search. 


^ Email address: daboyce@fs.fed. us 


296 


September 2005 


Techniques 


297 


Key Words: Northern Goshawk, Accipiter gentilis; detection rates; forest management; nest area; occupancy; 

repeated sampling 


<:CUANDO ESTA AUSENTE ACCIPITER GENTILIS? ^ES SUFICIENTE UNA SOLA VISITA PARA IN- 
FERIR AUSENCIA EN AREAS DE NIDIFICACION OCUPADAS? 

Resumen. — Probamos la eficiencia de tres metodos (busqueda de nidos historica, busqueda por medio 
reproduccion de grabaciones, busqueda a lo largo de transectos de arboles) para detectar la presencia 
del halcon Accipiter gentilis en areas de nidificacion activas durante la epoca reproductiva de 1994, 
utilizando una sola visita a un area de nidificacion previamente conocida. Utilizamos las tasas de detec- 
cion en un modelo de probabilidad para determinar cuantas visitas se requieren para tener certeza al 
reportar una ausencia de esta especie de halcon. El proposito de este estudio es entender si los tres 
metodos para detectar a esta especie son suficientemente robustos para confiar en ellos al tomar deci- 
siones de manejo de tierras que pueden afectar areas de nidificacion. Realizamos pruebas ciegas a traves 
del oeste de los Estados Unidos. Los resultados fueron similares entre los metodos; la presencia de los 
halcones no fue detectada en el 36-42% de las areas de nidificacion activas luego de una sola visita, 
Estos resultados indican que una sola visita a un area de nidificacion no es adecuada para obtener 
informacion confiable sobre la actividad de nidificacion en el area. Nuestro modelo de probabilidad de 
deteccion mostro que si cada metodo es repetido tres (historico o transecto de arboles) o cuatro (re- 
produccion de grabaciones) veces, la ausencia de halcones puede ser inferida con un alto grado de 
confianza. Las conclusiones con respecto a la actividad de las areas de nidificacion utilizando el metodo 
de muestreo de una sola visita deben tomarse con gran precaucion. La clasificacion de un sitio de 
nidificacion como vacio, cuando de hecho los halcones estan presentes, es una preocupacion seria y 
puede llevar a conclusiones falsas. Las personas encargadas de manejar las tierras y tomar decisiones 
con relacion a la alteracion de los habitats no deberian confiar en una sola visita a los sitios de nidifi- 
cacion para determinar la ausencia de estos halcones. Algunas de las posibilidades para mejorar la 
deteccion de halcones que se encuentran nidificando incluyen realizar visitas multiples e independientes 
utilizando la misma metodologia, utilizar conjuntamente una secuencia de tecnicas para producir me- 
jores probabilidades de deteccion acumulativas o desarrollar un metodo nuevo que pueda proveer de 
una probabilidad de deteccion mayor. La metodologia de busqueda de nidos historica obtuvo los me- 
jores resultados, seguida por la de los transectos de arboles y la busqueda por medio de reproduccion 
de grabaciones. 

[Traduccion del equipo editorial] 


The U.S. Department of Interior Fish and Wild- 
life Service (FWS) reviewed the status of the North- 
ern Goshawk {Accipiter gentilis atricapillus) for Fed- 
eral protection (i.e., listed as threatened or 
endangered under provisions of the Endangered 
Species Act) three times since 1991. In each case 
the FWS ruled that listing was unwarranted. Pop- 
ulation trend is one of five factors used by the FWS 
for determining whether to list a species as threat- 
ened or endangered. The majority of nesting gos- 
hawks in the western United States are located on 
lands managed by the U.S. Department of Agri- 
culture Forest Service (FS). Since the FS is re- 
quired by the National Forest Management Act 
(NFMA) to maintain species population viability, 
monitoring the occupancy of goshawk nest areas is 
necessary to evaluate population trends. 

Lacking a formal national goshawk monitoring 
program, the FS management approach to pro- 
tecting goshawks in the southwestern United States 


is to locate goshawk nest trees and post-fledging 
family areas (Kennedy et al. 1994) prior to habitat 
alterations and then to apply goshawk manage- 
ment recommendations (varying from region to 
region) to conserve the nest area, manage the 
post-fledging family area, and manage the foraging 
area (Reynolds et al. 1992) . After implementation 
of habitat management prescriptions, follow-up 
management practices should include monitoring 
the effect of habitat changes on species; however, 
this is rarely done. The untested assumption is that 
the management program will work as designed. 

Finding and monitoring nesting goshawks is a 
critical component of successful adaptive land 
management practices if goshawks are to persist in 
managed landscapes. Goshawks exhibit strong fi- 
delity to nest areas (Reynolds and Joy in press) , but 
have fluctuating population numbers and nesting 
success year to year. Goshawks also frequently 
change nest locations within a nest area or between 


298 


Boyce et al. 


VoL. 39, No. 3 


nest areas within a territory. Because a proportion 
of the local population of goshawks moves to al- 
ternate nest areas on an annual basis, sampling 
only the historical nest areas over time without 
finding the alternate nest areas will result in fewer 
and fewer occupied nest areas (i.e., the unwar- 
ranted appearance of a declining population) . 

Counting, sampling, and detecting birds are im- 
portant concerns of avian researchers (Bart and 
Earnst 2002, Farnsworth et al. 2002, Rosenstock et 
al. 2002, Thompson 2002). Developing techniques 
to find goshawks efficiently has been an ongoing 
process (Kimmel and Yahner 1990, Kennedy and 
Stahlecker 1993, Joy et al. 1994, Watson et al. 1999, 
Penteriani 1999, Roberson et al. in press). Biolo- 
gists have yet to develop an accurate, cost-effective 
method that will detect goshawks throughout the 
nesting period. This is because the species is secre- 
tive, difficult to find and study, and their behavior 
changes during the breeding season. Kennedy and 
Stahlecker (1993) tested a technique for broad- 
casting goshawk vocalizations from calling stations 
positioned on parallel transects that were placed 
tangential to the occupied nest. Their tests were 
conducted in the southwestern U.S. during the 
nestling to fledging stage. They found that the 
probability of detecting a goshawk, when within 
100 m of a nest, averaged 70% throughout the sea- 
son using multiple visits. The median detection dis- 
tance was 141 m. On control transects, without 
broadcasting, detection rates dropped to between 
30% (courtship) and 60% (fledgling). 

In Washington, Watson et al. (1999) tested Ken- 
nedy and Stahlecker’s (1993) broadcast method us- 
ing three stations (400 m, 250 m, and 100 m) on 
a single transect that passed tangential to the nest 
at 100 m at its closest point. They found five visits 
at 100 m from the nest, eight visits at 250 m from 
the nest, and 10 visits at 400 m attained a 90% or 
higher detection rate. In another study using the 
broadcast technique from courtship to fledgling 
dependency, only 52% of goshawks were detected 
(McClaren et al. 2003) ; but, detections were lower 
during courtship (40%) and highest during fledg- 
ling dependency (75%). Kennedy and Stahlecker 
(1993), Watson et al. (1999), and McClaren et al. 
(2003) are examples of experienced goshawk bi- 
ologists evaluating goshawk survey techniques. 
Their prior experience with goshawks and knowl- 
edge of nest locations may have positively influ- 
enced experimental results (i.e., their detection 


rates probably represent maximum rates under test 
conditions) . 

A problem with past goshawk inventory and 
monitoring efforts has been a reliance on meth- 
odologies whose bias, probability of detection, and 
magnitude of detection error were unknown. 
There has always been uncertainty associated with 
misclassifying a goshawk territory as unoccupied 
when it may be occupied (i.e., error of omission). 
In 1994, the FS identified the need to test the ef- 
ficacy of techniques for finding goshawks. This was 
driven by the FS desire to implement specific hab- 
itat altering management actions designed to pro- 
tect goshawk nest areas, post-fledging family areas, 
and the surrounding foraging area from harm 
(Reynolds et al. 1992), or to allow for flexible man- 
agement options if goshawks were not present. 
Three commonly used detection methods available 
at that time were identified as needing testing (his- 
torical nest tree search, broadcast search, and tree 
search within potential nest areas). No investiga- 
tors had compared the potential errors associated 
with the three typical inventory techniques. 

Our objectives were to: (1) document the error 
associated with each of these three detection tech- 
niques and (2) use the error rates to estimate the 
number of nest area visits needed to infer absence 
of goshawks with different levels of confidence. We 
conducted a blind test of these methods for de- 
tecting breeding goshawks to reveal the magnitude 
of error associated with each technique. The rea- 
son we conducted blind tests was to control the 
variability introduced in previous tests conducted 
by experienced goshawk biologists that had prior 
knowledge of the nest area and its status (Kennedy 
and Stahlecker 1993, Joy et al. 1994); possibly in- 
fluencing their results. We then input our results 
into a probability model to conceptually explore 
various combinations of detection rates, errors as- 
sociated with these detection rates, and predict the 
number of sampling visits needed to have confi- 
dence in the information collected. 

Methods 

We tested the efficacy of revisiting historical nest trees, 
broadcasting goshawk vocalizations in nest areas, and 
scanning all trees along transects established throughout 
nest areas within an 800 m diameter area centered on 
occupied (nest with eggs/young) nest areas. The size of 
our sampling unit was 1/35 the estimated size of the ter- 
ritory (2400 ha) (Reynolds et al. 1992) and was selected 
to account for alternate nest locations within a single nest 
area. Field tests were conducted from June to early mid- 
July 1994 during nestling and fledgling dependency pe- 


September 2005 


Techniques 


299 


riods (Squires and Reynolds 199'7). Experienced field bi- 
ologists determined that each nest area tested had 
nesting goshawks present prior to the test. During the 
testing period, occupancy was determined by observing 
goshawks incubating eggs, adults brooding young, or ot> 
serving young at the nest. The same criteria were used 
at all study areas. Personnel naive to the presence and 
location of occupied nests were used to test the three 
methods. Only one method was tested, and only one visit 
was made, at each occupied nest area. The three meth- 
ods were randomly assigned to active nest areas. To sim- 
ulate normal field conditions, experience was allowed to 
vary among field members; no effort was made to ran- 
domize field crew experience among the three detection 
methods. Results from each state were pooled to improve 
sample size. 

Study Areas. Tests were conducted in Arizona, Califor- 
nia, New Mexico, and Wyoming. In Arizona {N = 44), 
tests were conducted in the Apache/Sitgreaves, Coconi- 
no and Kaibab National Forests. Forests in Arizona were 
dominated by ponderosa pine {Pinus ponderosa ) , white fir 
{Abies concolor), and Douglas-fir {Pseudotsuga menziesii). In 
California {N =10), tests were conducted in the Klamath 
National Forest, where at higher elevations, forests were 
dominated by red fir {Abies magnified), white fir, ponde- 
rosa pine, lodgepole pine {Pinus contorta), Douglas-fir, 
and incense cedar {Calocedrus decurrens), and lower ele- 
vation forests by ponderosa pine and white fir (Kuchler 
1977). In New Mexico {N = 11) tests were conducted in 
the Santa Fe National Forest where forests contained 
ponderosa pine, Douglas-fir, white fir, and quaking aspen 
{Populus tremuloides) and at higher elevations subalpine 
fir {Abies lasiocarpa) and Englemann spruce {Picea engel- 
mannii). In Wyoming {N = 12), tests were conducted in 
the Medicine Bow National Forest where lower elevation 
forests contained lodge pole pine with scattered quaking 
aspen, and higher elevation forests contained subalpine 
fir and Engelmann spruce (Alexander et al. 1986, Mar- 
ston and Clarendon 1988). 

Historical Nest Search. The most common goshawk 
search technique used prior to 1990 was to visit historical 
nest areas and relocate previously used nest trees to de- 
termine occupancy. Typically, little effort was spent in a 
broader search of a nest area if goshawks were not found. 
To simulate this method, biologists were given 1:24000 
scale maps marked with the approximate locations of 
nest trees within a nesting area where goshawks had pre- 
viously nested. Biologists were instructed to relocate the 
nest trees and determine if goshawks were present and 
nesting. The strength of this method relies on goshawk 
fidelity to nest areas (Reynolds et al. 1994) and that field 
personnel often detect goshawk presence by observing 
the defensive behavior of goshawks near their nests. Oth- 
er clues to goshawk nest area occupancy with this method 
included observing fecal material, prey remains, or molt- 
ed goshawk feathers in the vicinity of nests. When these 
clues were found, the area was searched further to find 
the occupied nest. 

Broadcast Surveys. This goshawk detection technique 
was developed in the early 1990s and involved broadcast- 
ing taped goshawk calls (alarm and juvenile food beg- 
ging) to elicit a response. Field crews followed the pro- 
cedure of Kennedy and Stahlecker (1993), as modified 


by Joy et al. (1994). Recorded calls of goshawks were 
broadcast from stations located at 300-m intervals, on 
parallel transects, in an 800 m radius area. A search was 
initiated to locate visually the nest once a goshawk re- 
sponded. The broadcast method is a means of systemat- 
ically searching the landscape for goshawks. This method 
is also useful for locating nesting pairs that move to al- 
ternate areas within their territory. A problem with the 
technique is that goshawks do not always respond to the 
broadcast call when they are present, may respond with 
a silent approach, or may respond to broadcast calls 
when they are far away from their nest areas and, thus, 
confound results. Additional confounding factors in- 
clude seasonal effects and misidentification of calls such 
as Steller’s Jay {Cyanocitta stellen) mimicking goshawks 
(Kennedy and Stahlecker 1993). 

Tree Transect. The tree transect technique is a system- 
atic visual search of a forested area centered on the oc- 
cupied nest. This method involved field crews walking 
along parallel transects spaced 50 m apart while exam- 
ining individual trees along either side of and directly 
along the transect path for goshawk nests in tree crowns 
(Squires and Reynolds 1997). At 50 m, the probability of 
eliciting goshawk defensive behavior was assumed to be 
high because they could presumably hear or see the field 
crew. Crews also looked for prey plucking posts, fecal ma- 
terial or stains, and scattered prey remains that would 
provide evidence of a potential occupied nest nearby. 

The Model. To address our second objective, we input 
the estimates of detection obtained from each search 
method above into a probability model (McArdle 1990). 
This allowed an estimation of the sample size needed to 
have confidence that goshawks were absent. In other 
words, how many revisits to the nest area are necessary 
to conclude goshawks are absent? Guynn et al. (1985) 
and Reed (1996) used probability models to retrospec- 
tively estimate confidence in detecting a species. Kery 
(2002) applied their model prospectively to infer how 
many visits were needed to be statistically confident that 
the species being sampled was absent. 

McArdle’s (1990) probability model includes: (1) the 
number of sampling visits {N) to an area, (2) the species 
probability of detection (p) during any visit, (3) and con- 
fidence (a) levels acceptable to the investigator (usually 
95%, and therefore a = 0.05). Assuming all visits to gos- 
hawk nest areas are similar and independent, the prob- 
ability of not detecting nesting goshawks after N visits 
(Kery 2002) is: 

Probability =„=(!-„)» (n 

{N unsuccessful visits) 

We can solve for N and get: 

log (a) = NX log (1 - p) (2) 

N= log (a)/log (1 - p) (3) 

The minimum number of visits, needed to conclude 
that a 800-m radius circle containing a previously used 
nest area is unoccupied within a 95% confidence interval 
(a = 0.05) can be estimated by substituting the proba- 
bility of detection values (historical = 0.64, broadcast = 
0.58, transect = 0.62; see Results for details) into Equa- 
tion 4. 


300 


Boyce et al. 


VoL. 39, No. 3 


N^in = log (0.05) /log (1 - p) (4) 

Results 

The results were similar for each method tested; 
between 58-64% of the occupied nest areas were 
found (historical nest search [16/25], tree transect 
[16/26], broadcast surveys [15/26]). Conversely, 
between 36-42% of the occupied goshawk nest ar- 
eas were missed. The broadcast result for a single 
visit is identical to what Kennedy and Stahlecker 
(1993) reported. We did not test for temporal dif- 
ferences in the methods due to limited sample siz- 
es. Despite the poor performance of each method 
for detecting goshawks using a single visit to a nest 
area, each method may be repeated several times 
to increase the probability of detection (Kennedy 
and Stahlecker 1993, Watson et al. 1999, McClaren 
et al. 2003). Using the detection results, we esti- 
mated the number of visits (N^m) needed to infer 
goshawk absence at nest areas at the 95% confi- 
dence level (a = 0.05) as 2.9 for the historical nest 
search, 3.1 for the tree transect, and 3.5 for the 
broadcast survey. These detection results are only 
relevant to active nest areas. 

Increasing the confidence level while maintain- 
ing a consistent detection rate quickly increases the 
number of visits needed to infer goshawk absence 
at nest areas and renders the sampling effort un- 
realistic (Table 1). For example, if we set the con- 
fidence level to 0.95, and want to limit the number 
of visits to two, then the probability of detection 
required for a method to be effective would have 
to be nearly 80%. Given this scenario, the goal for 
developing new or improved detection techniques 
should be to achieve a probability of detection lev- 
el of at least 80%. If the confidence level is in- 
creased to 0.99 (a = 0.01) to further reduce the 
misclassification error while retaining the detection 
probability at 80%, then the number of required 


visits to nest areas is three and is still a feasible 
management option (i.e., not cost prohibitive). 
McKelvey and Pearson (2001) examined a series of 
simulations for measuring small mammal popula- 
tions with different detection probabilities and 
their results revealed the same general pattern as 
ours in that low detection probabilities require a 
large number of sampling sessions to attain confi- 
dence in the findings. 

Discussion 

Our results were from occupied nest areas only. 
Although we controlled as much variation as pos- 
sible, there were many sources of variation we did 
not control. We did not test for false positive de- 
tections at unoccupied sites (Kennedy and Stah- 
lecker 1993), which are needed for a broader de- 
scription of detection probabilities. Detection 
frequencies of goshawks at nest areas may vary for 
any number of reasons, but perhaps most impor- 
tant are changes in goshawk behavior as breeding 
season progresses (Squires and Reynolds 1997). 
Breeding goshawks become more defensive at nest 
areas later in the nesting season and generally are 
easier to detect (Squires and Reynolds 1997). 
Young goshawks also are easier to detect later in 
the breeding season as they grow and become 
more active (McClaren et al. 2003). Because detec- 
tion methods may be temporally sensitive, manag- 
ers must interpret the results cautiously (McClaren 
et al. 2003, Roberson et al. in press). 

As the breeding season progresses from March 
through July, goshawk nest failures continue for a 
host of reasons. A difficult sampling problem is to 
account for these nest failures. Sampling after re- 
productive failure occurs may lead to misclassifi- 
cation of nest areas as inactive. In addition, nesting 
areas are occupied by adults that do not breed ev- 


Table 1. Theoretical number of visits to Northern Goshawk nest areas to infer goshawk absence using different 
detection probabilities (p) and confidence levels (a). 


Probability of Detection 

a 

60 

65 

70 

75 

80 

85 

90 

95 

0.25 

1.51 

1.32 

1.15 

1.00 

0.86 

0.73 

0.62 

0.46 

0.20 

1.76 

1.53 

1.34 

1.16 

1.00 

0.85 

0.70 

0.54 

0 15 

2.07 

1.81 

1.58 

1.37 

1.18 

1.00 

0.82 

0.63 

0.10 

2.51 

2.19 

1.91 

1.66 

1.43 

1.21 

1.00 

0.77 

0.05 

3.26 

2.85 

2.49 

2.16 

1.86 

1.58 

1.30 

1.00 

0.01 

5.03 

4.39 

3.83 

3.32 

2.86 

2.43 

2.00 

1.54 


September 2005 


Techniques 


301 


ery year and thus, detection probabilities at indi- 
vidual nest areas are likely to vary temporally (Boal 
et al. 2005, R. Reynolds pers. comm.). We did not 
test the ability to detect nonbreeding pairs occu- 
pying nest areas; we tested only the breeding por- 
tion of the population (i.e,, actively nesting in 
pairs) . This has important ramifications for under- 
standing the population’s status (Kennedy 1997) 
and for managers making decisions based on re- 
sults for years when few pairs are breeding. The 
ability to detect nonbreeding goshawks and breed- 
ing goshawks that have failed are likely to be dif- 
ferent. Improved probabilities of detection may be 
possible by regulating the timing of when different 
methods are used. 

Another source of variation that affects detec- 
tion probabilities is the timing of egg-laying by fe- 
males within and between populations: variation in 
the timing of egg-laying introduces inherent error 
to detection rate estimates. Thus, there will likely 
be differential success in detecting goshawks be- 
cause the detection method used will not be per- 
fectly sequenced to the breeding phenology of all 
pairs within or between populations. We recom- 
mend that managers determine the breeding phe- 
nology of their target population before imple- 
menting goshawk surveys (see Dewey et al. 2003) . 

Variation also exists in the experience of field 
crews and therefore, accuracy and reliability of sur- 
vey data. In addition, goshawks may move to alter- 
nate nesting areas within a territory; this constant 
shifting among alternate nests may result in a per- 
ceived decay in the number of occupied nests and 
a fallacious conclusion of population decline if 
only the historical nest areas are visited (R. Reyn- 
olds pers. comm.). Given that multiple factors in- 
fluence detection probabilities, the implication for 
monitoring populations at regional scales is that 
detection protocols should consider these sources 
of variation so that data sets from different loca- 
tions and times are comparable for later use in an- 
alyzing large-scale population trends. 

None of the goshawk detection methods tested 
in this study, when applied once, were adequate to 
conclude goshawks were absent at nest areas. The 
usefulness of new detection methods is dependent 
on understanding the associated detection proba- 
bilities and error rates for different spatial and 
temporal scales. Future approaches might include 
combining several different methods in a temporal 
sequence that improves the cumulative probability 
of detection throughout the breeding season 


(Dewey et al. 2003). Highly accurate methods ap- 
propriate early in the breeding season (e.g., listen- 
ing stations; Dewey et al. 2003) may be ineffective 
late in the breeding period. However, by combin- 
ing methods and taking advantage of their 
strengths, improved results may be obtained, but 
this remains to be tested. Another approach is to 
test the detection probability of successive appli- 
cations of the historical and tree-transect methods 
(i.e., multiple visits) and determine if the results 
match the outcome reported for the broadcast 
method (70%; Kennedy and Stahlacker 1993). The 
predictions in this paper related to cumulative de- 
tection probabilities from multiple applications of 
one technique should be tested. If these predic- 
tions are supported empirically, then managers 
could design a monitoring program that relies on 
multiple applications of a single technique (e.g., 
tree transects). 

Detection success may be optimized by using lis- 
tening stations prior to egg-laying (March and 
April; Penteriani 1999, Dewey et al. 2003), tree 
searches on parallel transects during incubation 
and the nestling stage (May-June) , and broadcast 
calling (wail and food begging) during the post- 
fledging dependency period (Kennedy and Stah- 
lecker 1993, McClaren et al. 2003). Although 
broadcast surveys are frequently used during the 
nestling stage, recent tests of this approach by Rob- 
erson et al. (in press) in Minnesota suggest broad- 
cast surveys may not be an effective tool during this 
stage. Roberson et al. (in press) report high detec- 
tion rates with broadcast surveys during courtship 
(70%) and fledgling-dependency phases (68%). 
Detection rates were lowest during the nestling 
phase (28%), when there appeared to be higher 
variation in likelihood of detecting individuals. 

Acknowledgments 

We thank all the field assistants who participated in this 
study. We particularly thank Drs. Richard T. Reynolds, 
Winston Smith, Clayton M. White, and Curtis Flather for 
their review of the manuscript. Bob Nelson, former na- 
tional director of wil d life and fisheries management for 
the Forest Service, funded and encouraged this study. We 
also thank Clint Boal, Vincenzo Penteriani, and two 
anonymous reviewers for their helpful comments. 

Literature Cited 

Alexander, R.R., G.R. Hoffman, and J.M. Wirsing. 1986. 
Forest vegetation of the Medicine Bow National For- 
est in southeastern Wyoming: a habitat type classifi- 
cation. USDA, Forest Service Research Paper, RM-271, 


302 


Boyce et ai.. 


VoL. 39, No. 3 


Rocky Mountain Forest and Range Experiment Sta- 
tion, Fort Collins, CO U.S.A. 

Bart, J. and S. Earnst. 2002. Double sampling to esti- 
mate density and population trends in birds. Auk 119: 
36-45. 

Boat, C.W., D.E, Andersen, and RL. Kennedy. 2005. Pro- 
ductivity and mortality of Northern Goshawks in Min- 
nesota. J. Raptor Res. 39:222-228. 

Dewey, S.R., RL. Kennedy, and R.M. Stephens. 2003. Are 
dawn vocalization surveys effective for monitoring 
goshawk nest-area occupancy? J. Wildl. Manag. 67: 
390-397. 

Farnsworth, G.L., K.H. Pollock, J.D. Nichols, T.R. Si- 
mons, J.E. Hines, and J.R. Sauer. 2002. A removal 
model for estimating detection probabilities from 
point-count surveys. Auk 119:414—425. 

Guynn, D.C., Jr., R.L. Downing, and G.R. Askew. 1985. 
Estimating the probability of non-detection of low 
density populations. Cryptozoology 4:55-60. 

Joy, S.M., R.T. Reynolds, and D.G. Leslie. 1994. North- 
ern Goshawk broadcast surveys: hawk response vari- 
ables and survey cost. Pages 24-30 in Block, W.M., 
M.S. Morrison, and M.H Reiser (Eds.), The Northern 
Goshawk: ecology and management. Stud. Avian Biol. 
16:24-30. 

Kennedy, P.L. 1997. The Northern Goshawk {Accipiter gen- 
tilis atricapillus) : is there evidence of a population de- 
cline? Special issue on responses of forest raptors to 
management: a holarctic perspective, y. Raptor Res. 31: 
95-106. 

and D.W. Stahlecker. 1993. Responsiveness of 

nesting Northern Goshawks to taped broadcasts of 
three conspecific calls./. Wildl. Manag. 57:249-257. 

, J.M. Ward, G.A. Rinker, and J.A. Gessaman. 

1994. Post-Hedging areas in Northern Goshawk home 
ranges. Stud. Avian Biol. 16:75-82. 

Kery, M. 2002. Inferring the absence of a species-a case 
study of snakes./. Wildl. Manag. 66:330-338. 

Kimmel, J.T., AND R.H. Yahner. 1990. Response of North- 
ern Goshawks to taped conspecific and Great Horned 
Owl calls. / Raptor Res. 24:107-112. 

Kuchler, A.W. 1977. The map of the natural vegetation 
of California. Pages 909-938 m M.G. Barbour and J. 
Major (Ed-S.), Terrestrial vegetation of California. 
John Wiley, New York, NY. 

Marston, R.A., AND D.T. Clarendon. 1988. Land system 
inventory of the Medicine Bow Mountains and Sierra 
Madre Mountains, Medicine Bow National Forest, Wy- 
oming. USDA Forest Service Research Paper MBNF- 


88-01. Medicine Bow National Forest, Laramie, WY 
U.S.A. 

McArdle, B.H. 1990. When are rare species not there? 
Oikos 57:276-277. 

McClaren, E.L., P.L. Kennedy, and P.L. Chapman. 2003 
Efficacy of male goshawk food-delivery calls in broad- 
cast surveys on Vancouver Island. /. Raptor Res. 37: 
198-208. 

McKelvey, K.S. and D.E. Pearson. 2001. Population es- 
timation with sparse data: the role of estimators versus 
indices revisited. Canadian J. Zool. 79:1754—1765. 

Penteriani, V. 1999. Dawn and morning goshawk court- 
ship vocalizations as a method for detecting nest sites. 
/. Wildl. Manag. 63:511-516. 

Reed, J.M. 1996. Using statistical probability to increase 
confidence of inferring species extinction. Conserv 
Biol. 10:1283-1285. 

Reynolds, R.T, R.T. Graham, H.M. Reiser, R.L. Bassett, 
P.L. Kennedy, D.A. Boyce, Jr., G. Goodwin, R. Smith, 
AND E.L. Fisher. 1992. Management recommenda- 
tions for the Northern Goshawk in the southwestern 
United States. USDA For. Serv. Gen. Tech. Rep. RM- 
217. Rocky Mountain Research Station, Fort Collins, 
CO U.S.A. 

AND S.M. Joy. In Press. Demography of northern 

goshawks in northern Arizona, 1991-1996. Stud. Avi- 
an Biol.: In press. 

, D.G. Leslie, and S.M. Joy. 1994. Nest productiv- 
ity, fidelity, and spacing of Northern Goshawks in Ar- 
izona. Stud. Avian Biol. 16:106-113. 

Roberson, A.M., D.E. Andersen and P.L. Kennedy. In 
press. Effectiveness of broadcast surveys for Northern 
Goshawks: considerations of breeding phase, detec- 
tion distance, and effective area surveyed. /. Wildl. 
Manag.: In press. 

Rosenstock, S.S., D.R. Anderson, K.M. Giesen, T. Leu- 
KERING, and M.F. Carter. 2002. Landbird counting 
techniques: current practices and an alternative. Auk 
119:46-53. 

Squires, J.R. and R.T. Reynolds. 1997. The Northern 
Goshawk {Accipiter gentilis). In A. Poole and F. Gill, 
[Eds.], The birds of North America, No. 298. The 
Birds of North America, Inc., Philadelphia, PA U.S.A. 

Thompson, W.L. 2002. Towards reliable bird surveys: ac- 
counting for individuals present but not detected. Auk 
119:18-25. 

Watson, J.W., D.W Hays, and DJ. Pierce. 1999. Efficacy 
of Northern Goshawk broadcast surveys in Washing- 
ton State. J. Wildl. Manag. 63:98-106. 

Received 28 January 2004; accepted 15 May 2005 

Associate Editor: Clint W. Boal 


J Rapt&rRes. 39(3):303-309 
© 2005 The Raptor Research Foundation, Inc. 


QUANTIFYING NORTHERN GOSHAWK DIETS USING REMOTE 
CAMERAS AND OBSERVATIONS FROM BLINDS 


Andi S. Rogers^ 

Arizona Game and Fish Department, 3500 South Lake Mary Road, Flagstaff, AZ 86001 U.S.A. 

Stephen DeStefano 

U.S. Geological Survey, Massachusetts Cooperative Fish and Wildlife Research Unit, Holdsworth Natural Resource Center, 

University of Massachusetts, Amherst, MA 01003 U.S.A. 

Michael F. Ingraldi 

Arizona Game and Fish Department, Research Branch, 2221 West Greenway Road, Phoenix, AZ 85023 U.S.A. 

Abstract. — Raptor diet is most commonly measured indirectly, by analyzing castings and prey remains, 
or directly, by observing prey deliveries from blinds. Indirect methods are not only time consuming, 
but there is evidence to suggest these methods may overestimate certain prey taxa within raptor diet. 
Remote video surveillance systems have been developed to aid in monitoring and data collection, but 
their use in field situations can be challenging and is often untested. To investigate diet and prey delivery 
rates of Northern Goshawks {Accipiter gentilis) , we operated 10 remote camera systems at occupied nests 
during the breeding seasons of 1999 and 2000 in east-central Arizona. We collected 2458 hr of useable 
video and successfully identified 627 (93%) prey items at least to Class (Aves, Mammalia, or Reptilia). 

Of prey items identified to genus, we identified 344 (81%) mammals, 62 (15%) birds, and 16 (4%) 
reptiles. During camera operation, we also conducted observations from blinds at a subset of five nests 
to compare the relative efficiency and precision of both methods. Limited observations from blinds 
yielded fewer prey deliveries, and therefore, lower delivery rates (0.16 items/hr) than simultaneous 
video footage (0.28 items/hr) . Observations from blinds resulted in fewer prey identified to the genus 
and species levels, when compared to data collected by remote cameras. Cameras provided a detailed 
and close view of nests, allowed for simultaneous recording at multiple nests, decreased observer bias 
and fatigue, and provided a permanent archive of data. 

Key Words; Northern Goshawk, Accipiter gentilis; prey-delivery rate, diet, remote camera; video surveillance. 


CUANTIFICACION DE LA DIETA DE ACCIPITER GENTILIS UTILIZANDO CAMARAS DE VIDEO 
CON SISTEMA REMOTO DE VIGILANCIA Y OBSERVACIONES DESDE UN ESCONDITE 

Resumen. — Comunmente la dieta de las aves rapaces es medida indirectamente por medio de analisis 
de egagropilas y restos de presas, o directamente por medio de observaciones de entregas de presa 
desde un escondite de observacion. Los metodos indirectos no solo toman mucho mas tiempo sino que 
tambien existe evidencia que sugiere que estos metodos pueden sobre-estimar la importancia de ciertos 
taxa de presa en la dieta de las rapaces. Se han desarrollado sistemas remotos de vigilancia con camaras 
de video para ayudar con la observacion y la recoleccion de datos, pero su uso en situaciones de campo 
puede ser dificil y en muchos casos no es un metodo probado. Para investigar la dieta y las tasas de 
entrega de presa de Accipiter gentilis, utilizamos 10 sistemas de camaras remotas en nidos activos durante 
las epocas reproductivas de 1999 y 2000 en el centro oriente de Arizona. Recolectamos 2,458 horas de 
video util y logramos identificar 627 (93%) restos de presa hasta Clase (Aves, Mammalia o Reptilia). 
Entre los restos de presa identificados a nivel de genero, identificamos 344 (81%) mamiferos, 62 (15%) 
aves y 16 (4%) reptiles. Durante la operacion de las camaras tambien hicimos observaciones desde 
escondites de un subgrupo de cinco nidos para comparar la eficiencia relativa y precision de los dos 
metodos. Las observaciones limitadas desde escondites rindieron menos entregas de presa y por lo tanto 
rindieron tasas de entrega mas bajas que la documentada simultaneamente con camaras. Los datos 
obtenidos mediante observaciones desde escondites indicaron una habilidad reducida de este metodo 


^ Corresponding author’s email address: arogers@azfd.gov 


303 


304 


Rogers et al. 


VoL. 39, No. 3 


para identificar presas a nivel de genero y especie al ser comparados con los dates colectados de los 
videos de las camaras remotas. Las camaras produjeron una vista detallada y cercana de los nidos, 
permitieron la grabacion simultanea de varies nidos, redujeron el sesgo y la fatiga del observador y 
produjeron un archive permanente de dates. 

[Traduccion del Equipo Editorial] 


Information on diet is important in understand- 
ing aspects of avian ecology such as diet overlap 
among species, predation, and prey selection (Ro- 
senberg and Cooper 1990, Redpath et al. 2001). 
Diet assessment in raptors is usually done indirect- 
ly, by recovering pellets and prey remains, or di- 
rectly, by observing prey deliveries from blinds; 
however, accurate estimates of raptor diet can vary 
depending on the technique employed (Marti 
1987). 

Indirect diet assessment can provide quantitative 
and qualitative information because raptors often 
leave behind undigested remnants of bones, feath- 
ers, and keratinous material as pellets, or as prey 
remains (Reynolds and Meslow 1984, MacLaren et 
al. 1988, Steenhof and Kochert 1988, Boal and 
Mannan 1994). However, prey remains and pellets 
may bias the representation of certain prey items 
(e.g., bird feathers are more easily detected than 
small bones); therefore, avian prey may be over- 
represented in raptor diet (Simmons et al. 1991, 
Bielefeldt et al. 1992). Marti (1987) suggested that 
pellet analysis is accurate only for raptor species 
that swallow their prey whole. Loss of prey remains 
to scavengers, investigator disturbance in the nest- 
ing area, and miscounting of remnant and incom- 
plete remains may also bias or limit results. 

Direct observation of raptors is a more accurate 
method for investigating diet in species that do not 
swallow their prey whole. Observations can be 
made from a blind within the nesting area; how- 
ever, observations near nests can disturb hawks, are 
labor intensive and require dawn to dusk obser- 
vations to obtain complete samples. In addition, 
direct observation requires positioning of the blind 
so that a view inside the nest bowl is possible (Col- 
lopy 1983). 

A more recent technology for studying diet in- 
volves remote cameras at raptor nests (Ouchley et 
al. 1994, Booms and Fuller 2003, Lewis 2004a). Ad- 
vantages of video surveillance for measuring diet 
include a reduction in observer bias and fatigue, 
minimal impact on an animal’s behavior, detailed 
information on diet composition, and an archival 
record of footage (Kristan et al. 1996, Stewart et 
al. 1997, Delaney et al. 1999). 


Lewis et al. (2004b) compared three methods for 
assessing raptor diet: video recording, pellet anal- 
ysis, and prey remain analysis. They found that 
quantifying prey using either prey remains or pel- 
let analysis did not provide as complete a descrip- 
tion of diet when compared to remote cameras. 
They did not, however, compare observations from 
blinds to remote cameras. In this paper, we de- 
scribe a camera system, monitoring, and data col- 
lection using remote video technology, and discuss 
advantages and disadvantages. In addition, we con- 
ducted limited observations from blinds at five 
nests and simultaneously collected data with re- 
mote cameras to compare the two methods. 

Study Area 

We conducted this study on the Sitgreaves portion of 
the Apache-Sitgreaves National Forest in east-central Ar- 
izona. The Sitgreaves portion encompasses ca. 350 800 ha 
(elevation = 1768-2417 m) and is located atop the Mo- 
gollon Rim on the southern edge of the Colorado Pla- 
teau. The Mogollon Rim is a large glacial escarpment 
that extends east across central Arizona into New Mexico. 
The Mogollon Rim edge has deep drainages with mixed- 
conifer communities of Douglas-fir {Pseudotsuga menzie- 
sii), white fir {Abies concolor), trembling aspen (Populus 
tremuloides) , ponderosa pine {Pinus ponderosa) , New Mex- 
ican locust {Robinia neomexicana) , and Gambel oak (Quer- 
cus gambelii; Brown 1994). Ridgetops are dominated by 
ponderosa pine forest. 

Methods 

We chose video monitoring as the primary method to 
quantify the diet of breeding Northern Goshawks {Acap- 
iter gentilis) in east-central Arizona (Rogers et al. in press). 
During the breeding seasons of 1999 and 2000, we ran- 
domly selected 10 nests (four in 1999 and six in 2000) 
from a pool of known territories {N =48). During June 
1999 and 2000, we mounted EOD-1000 Electro-optics® 
remote cameras (Electro-Optics, St, Louis, MO U.S.A.) 
when nestlings were between 4-7 d old (nestlings were 
shaded during camera installation). Cameras ran from 22 
June-18 July 1999 and 6 June-31 July 2000. We needed 
a minimum of three people for camera placement with 
a mean setup time of 110 min per nest (range = 80-132 
min). Nest trees were ponderosa pine or Douglas-6r, and 
nest heights were ca. 20 m above ground. 

Cameras were 3.5 X 12 cm and equipped with 3.6 mm 
lenses. Each camera had 380 lines of resolution and a 
one-lux digital color system. During installation, the 
ground crew viewed the nest using a Broksonic D.C. TV/ 
VCR combination (Broksonic Corporation of America, 


September 2005 


Techniques 


305 



Figure 1. Schematic of camera system for monitoring 
Northern Goshawk nests in east-central Arizona in 1999 
and 2000. 


Table 1. Cost of video surveillance equipment used for 
a diet study of Northern Goshawks in Arizona during 
breeding seasons 1999 and 2000. Prices based on 1999 
retail costs associated with assembly of one system. 


Component 

Approximate Cost 
($ US) 

VHS time-lapse recorder® 

675 

Remote video camera 

250 

DC television monitor 

190 

Rechargeable battery 

180 

2-amp battery charger‘s 

85 

Coaxial and power cables and 
connectors 

80 

50 caliber ammunition can/locks 
and cables 

50 

Total 

1470 


“ Cost of Panasonic and Sony VHS recorder were averaged (Pan- 
asonic = $810. 00/Sony = $520.00). 

^ TV/VCR combo used for multiple units. 


New York, NY U.S.A.) while a person in the nest tree 
positioned the camera. Once positioned, we secured the 
camera to the trunk of the tree or an overhanging 
branch. The goal in camera placement was a field of view 
that contained the entire nest structure and focused on 
the nest bowl. This was achieved by positioning cameras 
about 3 m away from nests at about a 45° angle to the 
nest structure (Fig. 1 ) . Cameras were connected to 75 m 
of durable telephone-power cord and coaxial video cable 
(copper coated RG-59) tacked along the trunk of the 
tree. Camera cords were attached to a Panasonic® AG- 
1070 DC (Panasonic, Secaucus, NJ U.S.A.) or Sony® SVT- 
DL224 (Sony, Park Ridge, NJ U.S.A.) time-lapse VHS re- 
corder, which were placed at least 50 m from the nest 
tree. Both time-lapse VCR models were industrial grade, 
12-volt DC models with time-lapse programming capabil- 
ity. VCRs were housed in military ammunition cans (20 
mm) and powered by one 12-volt, 64 amp-hr sealed Op- 
tima® (Optima, Denver, CO U.S.A.) rechargeable lead 
acid battery (22 kg each). Batteries were kept dry under 
a plastic bin. Ammunition cans were locked, and all 
ground equipment was secured to a tree. Finally, we cov- 
ered equipment with forest litter for shade and camou- 
flage. Cost for one complete system was about $1470 US 
(Table 1). 

We programmed VCRs to record 5 frames/ sec, which 
provided up to 24 hr of footage per videotape. We re- 
corded activity at each nest in a 2-d sequence (12 hr/d), 
and cameras recorded 6 of 7 d of the week. We recorded 
from 0450-1650 H on day one and 0800-2000 H on day 
two. During 1999, all batteries and tapes were changed 
at night to reduce disturbance to hawks. In 2000, battery 
and tape changes occasionally occurred before nightfall, 
but only if ground equipment was located out of sight of 
nests. No more than 5 min were spent within nest stands 
changing batteries and tapes every other night or day. 
We continued to record video at nests until fledgling 
Northern Goshawks did not receive prey deliveries for 2 
consecutive days. 


We viewed video footage on a 19 in Toshiba® television 
(Toshiba, Irvine, GA U.S.A.) with a JVC® SuperVHS VCR 
(JVC, Wayne, NJ U.S.A.). Prey items were identified to 
Class (Mammalia, Aves, or Reptilia), genus and species 
when possible, or classified as unknown. Prey items iden- 
tified to class only were characterized as small, medium, 
or large based on a priori size class categories from Cock- 
rum and Petryszyn (1992) and Dunning (1993). 

To compare methods, we observed goshawks (2000 
breeding season) at a subset of five nests from blinds 
constructed 25—40 m from nests. We erected blinds on 
ground prior to sunrise before each observation period. 
Blinds were constructed of camouflage heavy-duty canvas 
with screen vtindows in all directions, which allowed for 
observation of hawks within the immediate nesting area. 
We used 8 X 32 binoculars and a 20 X 60 spotting scope 
to count and identify prey items delivered to nests. Items 
were identified to Class (Mammalia, Aves, or Reptilia) 
and to genus and species when possible. We initiated 
blind observations when nestlings were between 8-12 d 
old and continued observing prey deliveries until young 
fledged. We conducted observations in 3-4 hr blocks 
each day starting at sunrise, and blinds were disassembled 
upon completion of each observation period. We allowed 
a 20-min acclimation period for adults and young before 
beginning observation. Observations from blinds were 
done in conjunction with video monitoring to compare 
accuracy of the two methods. In addition, observations 
from blinds and video reviewing were done by one per- 
son (Rogers) to minimize bias. 

Results 

Camera Results. Adult Northern Goshawks ac- 
tively defended the nest while we placed cameras 
in nest trees. However, adult females returned to 
the nest within 10-20 min after we vacated terri- 



306 


Rogers et al. 


VoL. 39, No. 3 


tories, as documented from video. In addition, we 
had no nest abandonment due to camera pres- 
ence, and eight of 10 nests were successful (i.e., 
fledged young). One nesting attempt failed 
due to adult female mortality (Bloxton et al. 2002) , 
and one was depredated by a Great Horned Owl 
{Bubo virginianus) . Adults did not flush during 
nighttime battery and tape changes. Adults flushed 
infrequently during daytime changes because we 
were out of sight of the nest tree. 

We collected 2458 hr of usable video footage, 
and about 500 hr were spent viewing tapes to iden- 
tify and quantify prey items. Approximately 50 hr 
were spent changing batteries and tapes, excluding 
travel time. We documented 676 prey deliveries 
from camera footage. Of these, we identified 627 
(93%) prey items to Class (Aves, Mammalia, or 
Reptilia) and observed a mean delivery rate of 0.28 
(SE = 0.02) prey items/hr. We were able to iden- 
tify, at least to genus, 422 (62%) of all prey items. 
Of items identified to at least genus, 344 (81%) 
were mammals, 62 (15%) were birds, and 16 (4%) 
were lizards. 

Direct Observations. Because blinds were con- 
structed before sunrise, adults infrequently flushed 
from nests. However, during disassembly and exit- 
ing nest territories, adults actively defended nests. 
When adults did flush from nests prior to an ob- 
servation period, they returned to the nest within 
about 10 min. We observed goshawks for a total of 
43 hr at five nests. We viewed seven prey deliveries, 
all of which were identifiable to Class. Mean prey 
delivery rate observed from blinds was 0.16 (SE = 
0.06) items/hr. 

Camera Versus Direct Observation. Camera foot- 
age yielded a higher total number of prey items 
delivered; therefore, our estimated prey delivery 
rate derived from video footage was higher than 
that derived from direct observation. An important 
result was that the camera footage revealed 12 de- 
liveries during our observation period in which we 
visually documented only seven deliveries. Accura- 
cy of prey identification to class was 100% using 
both methods, but we were able to document 58% 
of all prey to genus and/or species from the video 
footage compared to 0% from direct observations. 

Discussion 

Use of remote camera systems is becoming a 
popular technique in wildlife studies, especially as 
equipment costs decrease. Eor example, the video 
surveillance equipment used in Lewis et al. 


(2004a) was over $2000 US and did not include 
batteries and chargers. Our equipment was similar 
to Lewis’s, but the cost was $1470 US, which in- 
cluded batteries and chargers. 

Cameras have been used to monitor diet, pre- 
dation events, and various behaviors of many spe- 
cies of wildlife (e.g., Wisniewski 1983, Sykes et al. 
1995, Hughes and Shorrock 1998, King et al. 
2001). Responses to camera installation may vary 
by species and individuals, timing of camera place- 
ment during the nesting season, and length of 
time needed for camera installation. Several work- 
ers reported no sign of nest abandonment due to 
cameras (Estes and Mannan 2003, Booms and Ful- 
ler 2003, Lewis et al. 2004a). However, Cain (1985) 
reported nest abandonment by Bald Eagles (Hal- 
iaeetus leucocephalus) due to camera installation. 
During our study, goshawks were distressed when 
cameras were installed, but did not seem to be af- 
fected by camera presence. In videos, adult and 
juvenile goshawks occasionally could be seen look- 
ing up at cameras, and there were several occasions 
when hawks perched directly below or on cameras. 
Goshawks were also distressed during our obser- 
vations from blinds, especially during our exit from 
territories, which suggested that direct observa- 
tions were more stressful to the nesting hawks than 
use of cameras. 

Remote cameras can facilitate sampling for ex- 
tended periods of time with a reduction in observ- 
er bias (Delaney et al. 1999) . In contrast, observa- 
tions from blinds are often done by more than one 
person, which increases the risk of observer bias 
(Boal and Mannan 1994). Video monitoring can 
also increase daily coverage because unmanned 
units can operate continuously. For example, we 
were able to collect nearly 2500 hr of observations 
in 2 yr with 10 cameras, whereas Boal (1993) and 
Boal and Mannan (1994) collected 1500 hr of ob- 
servation from blinds, which required three field 
assistants per year for 3 yr. Observer fatigue could 
also bias the results based on direct observation, 
and the cost of labor would be high. 

Using remote cameras, we were able to record 
prey deliveries at the nest for up to 1 mo after 
Northern Goshawk young fledged (Rogers et al. in 
press). In contrast, observations from blinds and 
pellet and prey remains collection are often dis- 
continued shortly after young fledge (MacLaren et 
al. 1988, Seguin et al. 1998). Although prey deliv- 
ered to nests during branching and fledgling stag- 
es often occurred out of camera range, allowing 


September 2005 


Techniques 


307 



Figure 2. Images of Northern Goshawk nests taken from video footage in east-central Arizona: (a) Female goshawk 
feeding 5-d-old young, (b) Golden-mantled ground squirrel and plucked Stellar’s Jay in the nest with 25-30 d old 
goshawk young. 


308 


Rogers et al. 


VoL. 39, No. 3 


cameras to operate longer provided additional 
qualitative information on post-fledgling diet. Ad- 
ditional advantages of cameras include decreased 
frequencies of observer entrances and exits within 
territories. We spent no more than 5 min in nest 
stands every other day and usually did not flush 
adults from nests. We strongly recommend chang- 
ing batteries and tapes at night, or alternatively lo- 
cating ground equipment 50 m or more from nest 
trees. 

Most importantly, using remote cameras greatly 
increased our ability^ to identify genus and species 
of prey delivered to nests. With cameras, we were 
able to see shape and color of most prey items (Fig. 
2). Mammals were easiest to identify to genus and 
species due to their size and distinctive pelage, as 
well as the ability to see feet and tails. Small birds 
were the most difficult to identify, but the ability 
to see feathers, and hence make an identification, 
distinctly increased if the adults plucked avian prey 
in the nest. 

Our data indicated that observations from blinds 
resulted in underestimates of prey numbers and 
delivery rates, but this needs further investigation. 
Prey items were missed in two ways during obser- 
vations from blinds. First, on some occasions we 
failed to notice small prey items brought by the 
female because we were focusing on identifying an 
item brought previously by the male. Second, we 
missed some prey items that were delivered to nests 
after dark or prior to daybreak. We did not use 
these items in calculating prey delivery rates, but 
included them in total prey deliveries. Without the 
ability to play back the videotape, we would not 
have noticed these prey items. 

A final advantage of video monitoring was the 
ability to record infrequent behavioral events. For 
example, during the 1999 nesting season, we doc- 
umented an attempted predation by a Red-tailed 
Hawk (Buteo jamaicensis), and in 2000, we recorded 
a bobcat (Lynx rufus) scavenging prey from a nest 
that had already fledged young. 

There are some limitations and constraints to us- 
ing camera systems. We experienced technical dif- 
ficulties including rodent and ungulate damage to 
cords, battery failure, loose connections, and water 
damage. In addition, when cameras’ angles were 
>45° to the nest, the view was often obstructed by 
the adult female’s back. To alleviate this problem, 
we searched for alternative branches or nearby 
trees that allowed for a 45° angle to the nest bowl. 
We recommend placing cameras opposite the di- 


rection of the adult flight pathway to the nest. 
Therefore, observing adult movement patterns near 
nests before camera placement is recommended. 
Video monitoring involves a relatively high initial 
cost. Also, as of 2000, no audio capability was avail- 
able within a waterproof system. Thus, collecting 
data on vocalizations during prey deliveries was not 
possible. One additional disadvantage of video tech- 
nology is the additional time required to transcribe 
video data. Even though tapes were fast forwarded 
during non-prey delivery times, it took ca. 1 hr of 
viewing to transcribe data for every 5 hr of video 
footage collected. We suggest viewing collected tap- 
es daily to minimize backlog and to allow research- 
ers to become aware of system problems before data 
collection is complete. 

In conclusion, we think remote cameras allowed 
us to collect more accurate diet data than if we 
would have solely used blind observations. Camera- 
monitoring systems are efficient, relatively nonin- 
vasive tools for quantifying diet and behavior of 
raptors. 

Acknowledgments 

This study was funded by the Arizona Game and Fish 
Department and the Arizona Cooperative Fish and Wild- 
life Research Unit at the University of Arizona. We thank 
Patricia Kennedy, Michael Collopy, and Robert Lehman 
for their earlier reviews of this manuscript. We also thank 
the numerous field assistants associated with this research 
and Susan MacVean for the Spanish translation of the 
abstract. 

Literature Cited 

Bielefeldt, J., R.N. Rosenfield, andJ.M. Papp. 1992. Un- 
founded assumptions about diet of the Cooper’s 
Hawk. Condor 94:427—439. 

Bloxton, T.D., A.S. Rogers, M.F. Ingraldi, S. Rosen- 
stock, J.M. Marzluff, and S. Finn. 2002. Possible 
choking mortalities of adult Northern Goshawks. J 
Raptor Res. 36:141-143. 

Boat, C.W. and R.W. Mannan. 1994. Northern Goshawk 
diets in ponderosa pine forests on the Kaibab plateau. 
Stud. Avian Biol. 16:97-102. 

Booms, T.L. and M.R. Fuller. 2003. Time-lapse video sys- 
tem used to study nesting Gyrfalcons. f. Field Ornithol 
74:416-422. 

Brown, D.E. 1994. Biotic communities: southwestern 
United States and northwestern Mexico. University of 
Utah Press, Salt Lake City, UT U.S.A. 

Cain, S.L. 1985. Nesting activity time budgets of Bald Ea- 
gles in southeast Alaska. M.S. thesis, Univ. Montana, 
Missoula, MT U.S.A. 

CocKRUM, E.L. and Y. Petrvszyn. 1992. Mammals of the 
southwestern United States and northwestern Mexico 
Treasure Chest Publications, Tucson, AZ U.S.A. 


September 2005 


Techniques 


309 


COLLOPY, M.W. 1983. A comparison of direct observa- 
tions and collections of prey remains in determining 
the diet of Golden Eagles. J. Wildl. Manag. 47:360- 
368. 

Delaney, D.K., T.G. Grubb, and D.K. Garcelon. 1999. 
An infrared video camera system for monitoring di- 
urnal and nocturnal raptors./. Raptor Res. 32:290-296. 

Dunning, J.B., (Ed.). 1993. CRC handbook of avian body 
masses. CRC Press Inc., Boca Raton, FL U.S.A. 

Estes, W.A. and R.W. Mannan. 2003. Feeding behavior 
of Cooper’s Hawks at urban and rural nests in south- 
eastern Arizona. Condor 105:107—116. 

Hughes, A.G. and G. Shorrock. 1998. Design of a du- 
rable event detector and automated video surveil- 
lance unit. /. Field Ornithol. 69:549—556. 

King, D.I., R.M. DeGraaf, RJ. Champlin, and T.B. Cham- 
PLIN. 2001. A new method for wireless video monitor- 
ing of bird nests. Wildl. Soc. Bull. 29:349—353. 

Kristan, D.M., R.T. Golightly, Jr., and S.M. Tomkiew- 
ICZ, Jr. 1996. A solar-powered transmitting video cam- 
era for monitoring raptor nests. Wildl. Soc. Bull. 24: 
284-290. 

Lewis, S.B., P. DeSimone, M.R. Fuller, and K. Titus. 
2004a. A video surveillance system for monitoring 
raptor nests in a temperate rainforest environment. 
Northwest Sci. 78:70-74. 

, M.R. Fuller, and K. Titus. 2004b. A comparison 

of 3 methods for assessing raptor diet during the 
breeding season. Wildl. Soc. Bull. 32:373-385. 

MacLaren, P.A., S.H. Anderson, and D.E. Runde. 1988. 
Food habits and nest characteristics of breeding rap- 
tors in southwestern Wyoming. Great Basin Nat. 48: 
548-553. 

Marti, C.D. 1987. Raptor food habits studies. Pages 67- 
80 in B.A. Giron Pendleton, B.A. Millsap, K.W. Cline, 
and D.M. Bird [Eds.], Raptor management tech- 
niques manual. National Wildlife Federation, Wash- 
ington, DC U.S.A. 

OucHLEY, K., R.B. Hamilton, and S. Wilson. 1994. Nest 


monitoring using a micro-video camera. J. Field Orni- 
thol. 65:410-412. 

Redpath, S.M., R. Clarke, M. Madders, and SJ. Thir- 
GOOD. 2001. Assessing raptor diet comparing pellets, 
prey remains, and observational data at hen harrier 
nests. Cowdor 103:184-188. 

Reynolds, R.T. and E.C. Meslow. 1984. Partitioning of 
food and niche characteristics of coexisting Accipiter 
during breeding. Auk 101:761-779. 

Rogers, A.S., S. DeStefano, and M.F. Ingraldi. 2005. 
Diet, prey delivery rates, and prey biomass of North- 
ern Goshawks in east-central Arizona. Stud. Avian Biol. 
31: In press. 

Rosenberg, K.V. and R.J. Cooper. 1990. Approaches to 
avian diet analysis. Stud. Avian Biol. 13:80-90. 

Seguin, J.F., P. Bayle, J.C. Thibault, J. Torre, and J.D. 
Vigne. 1998. A comparison of methods to evaluate the 
diet of Golden Eagles in Corsica./. Raptor Res. 32:314— 
318. 

Simmons, R.E., D.M. Avery, and G. Avery. 1991. Biases 
in diets determined from pellets and remains: correc- 
tion factors for a mammal and a bird-eating raptor./. 
Raptor Res. 25:63-67. 

Steenhof, K. and M.N. Kochert. 1988. Dietary respons- 
es of three raptor species to changing prey densities 
in a natural environment./. Anim. Ecol. 57:37-46. 

Stewart, P.D., S.A. Ellwood, and D.W. MacDonald. 
1997. Remote video-surveillance of wildlife — an intro- 
duction from experience with the European badger 
Meles meles. Mammal Review 27:185-204. 

Sykes, P.W., W.E. Ryman, C.B. Kepler, and J.W. Hardy. 
1995. A 2-hour remote surveillance system for terres- 
trial wildlife studies./. Field Ornithol. 66:199-211. 

Wisniewski, L. 1983. Remote video monitoring of eagles 
and falcons. Pages 58—61 m T.M. Ingram [Ed.], Bald 
Eagle restoration. Eagle Valley Environmentalists, Ap- 
ple River, IL U.S.A. 

Received 26 March 2004; accepted 15 December 2004 

Guest Editor: Patricia L. Kennedy 


/ Raptor Res. 39(3) :310-323 
© 2005 The Raptor Research Foundation, Inc. 


TEMPORAL PATTERNS OF NORTHERN GOSHAWK NEST AREA 
OCCUPANCY AND HABITAT: A RETROSPECTIVE ANALYSIS 

Steven M. Desimone ^ and Stephen DeStefano^ 

U.S. Geological Survey, Oregon Cooperative Wildlife Research Unit, 104 Nash Hall, Oregon State University, 

Corvallis, OR 97331 U.S. A 

Abstract. — We studied occupancy and habitat associations of Northern Goshawks {Accipiter gentilis) at 
nest areas in south-central Oregon in 1992-94. We surveyed 51 pre-1992 nest areas (i.e., historical 
breeding areas first discovered during 1973-91) for goshawks and used aerial-photograph interpretation 
to document forest cover conditions and changes over time between areas that were occupied by gos- 
hawks and those where we did not detect goshawks (no-response sites). We also surveyed for new nests 
during 1992-94. Of 38 occupied nests first found in 1992-94 (i.e., post-1992 nest areas), 86% (33/38) 
were in mid-aged (mean stand DBH 23-53 cm, <15 trees/ha >53 cm DBH) or late (>15 trees/ha >53 
cm DBH; mean stand DBH >53 cm) closed (>50% canopy closure) structural-stage forest. Occupancy 
of historical (pre-1992) nest areas by goshawks was 29% (15/51). Of 46 pre-1992 nest areas that we 
examined for habitat change, 15 were occupied by goshawks in 1994 and had more mid-aged closed 
and late closed forest in 12-, 2T, 52-, 120-, and 1 70-ha circular areas centered on nest locations than 
did 31 no-response areas. There was no difference in the amount of late closed and mid-aged closed 
forest in pre-1992 nest areas compared with occupied pre-1992 nest areas. A logistic regression model 
for all occupied nest areas confirmed that late closed and mid-aged closed forest variables were impor- 
tant indicators of forest conditions that supported breeding pairs. Goshawks were more likely to persist 
in the historical nest areas that had about 50% of mature and older closed-canopy forest within the 52- 
ha scale. We recommend retaining existing late closed, late open, and mid closed structure within 52- 
ha scale of the nest site. Moreover, late closed and mid closed structure combined should not fall below 
50% within the 52-ha scale and should exceed 40% within the 1 70-ha scale surrounding the nest site. 

To optimize conditions for breeding goshawks, we recommend retaining large trees (>53 cm DBH) to 
help preserve stand integrity, maintain closed canopies, and provide connectivity to alternative nest sites 
within nest areas. 

Key Words; Northern Goshawk, Accipiter gentilis; habitat, historical nest areas', landscape change, Oregon. 


PATRONES TEMPORALES DE OCUPACION DE areas DE NIDIEICACION Y HABITAT DE ACCIP- 
ITER GENTILIS: UN ANALISIS RETROSPECTD/O 

Resumen. — Estudiamos la ocupacion y las asociaciones de habitat de Accipiter gentilis en areas de nidifi- 
cacion del centro-sur de Oregon entre 1992 y 1994. Tambien censamos 15 areas de cria historicas 
descubiertas entre 1973 y 1991 (i.e., nidificacion pre-1992), y usamos fotografias aereas para documen tar 
las condiciones de cobertura de bosque y cambios en el tiempo entre areas que estaban ocupadas por 
esta especie y areas en las que no la detectamos (sitios sin respuesta). Tambien realizamos censos para 
buscar nidos nuevos entre 1992 y 1994. De 38 nidos activos encontrados por primera vez entre 1992 y 
1994 (i.e. nidificacion post-1992), el 86% (33/38) se encontro en bosques de sucesion media (promedio 
de DAP 23-53 cm, <15 arboles/ha >53 cm DAP) o bosques cerrados antiguos (>15 arboles/ha >53 
cm DAP; promedio de DAP >53 cm; >50% de cobertura del dosel). La ocupacion de las areas de 
nidificacion historicas (pre-1992) por parte de A. gentilis ine del 29% (15/51). De 46 sitios de crfa pre- 
1992 para los cuales evaluamos los cambios en el habitat, 15 estuvieron ocupados en 1994 y presentaron 
mayor cantidad de bosques cerrados de edad media y bosques antiguos en areas circulares de 12, 24, 
52, 120 y 170 ha centradas en sitios en donde se ubicaban nidos, que 31 sitios sin respuesta. No existio 


^ Corresponding author’s present address; Washington Department of Fish and Wildlife, 600 Capitol Way North, 
Olympia, WA 98501-1091 U.S.A.; email address; desimsrad@dfw.wa.gov 

^ Present address: U.S.G.S. Massachusetts Cooperative Fish and Wildlife Research Unit, Holdsworth Natural Resources 
Center, University of Massachusetts, Amherst, MA 01003 U.S.A. 


310 


September 2005 


Conservation 


311 


diferencia en la cantidad de bosques cerrados de edad media y bosques antiguos entre areas de nidi- 
ficacion pre-1992 en comparacion con las areas ocupadas pre-1992. Un modelo de regresion logistica 
que incluyo todas las areas de nidificacion ocupadas confirmo que las variables de los bosques cerrados 
de edad media y sucesion tardia fueron indicadoras importantes de las condiciones del bosque propicias 
para las parejas reproductivas. Las aves presentaron mayor probabilidad de persistir en las areas de 
nidificacion historica que presentaban aproximadamente el 50% de bosques maduros antiguos de dosel 
cerrado a la escala de 52 ha. Recomendamos mantener la estructura de bosques antiguos cerrados y 
abiertos y bosques de edad media cerrados en las 52 ha circundantes a los sitios de nidificacion. Ademas, 
la estructura combinada de bosques cerrados antiguos y de edad media no debe caer por debajo del 
50% a la escala de 52 ha y no debe exceder el 40% en la escala de las 170 ha circundantes a los sitios 
de nidificacion. Para optimizar las condiciones para la nidificacion de A. gentilis, recomendamos man- 
tener arboles grandes (>53 cm DAP) para ayudar a preservar la integridad de los bosques, mantener 
doseles cerrados y proveer conectividad entre sitios de nidificacion alternatives ubicados en las mismas 
areas de cria. 

[Traduccion del equipo editorial] 


The ability of breeding pairs of Northern Gos- 
hawks (Accipiter gentilis; hereafter, goshawks) to per- 
sist in intensively managed and selectively harvest- 
ed forests over time is largely unknown. Evidence 
suggests tree harvest impacts nest site selection 
(Crocker-Bedford 1990, Penteriani and Faivre 
2001), use (Woodbridge and Detrich 1994), and 
ultimately nesting persistence (Crocker-Bedford 
1995). Penteriani and Faivre (2001) examined log- 
ging disturbance and habitat change over a limited 
time (6-11 yr) in a European shelterwood harvest 
regime, but the effects of habitat alteration in west- 
ern North American forests are not fully under- 
stood. Mature forest, consisting of large trees (di- 
ameter at breast height [DBH] >50 cm) and 
closed canopy cover (>50%), was demonstrated to 
be preferred by breeding goshawks for nest sites in 
western North America (e.g., Hayward and Escano 
1989, Bull and Hohmann 1994, Squires and Rug- 
giero 1996, Daw and DeStefano 2001, McGrath et 
al. 2003). 

There has been concern and debate that gos- 
hawk populations in western North America may 
be declining in response to habitat alteration and 
loss of these forests (Kennedy 1997, DeStefano 
1998, Smallwood 1998, Crocker-Bedford 1998). Us- 
ing aerial photographic records of timber harvest 
areas (Reutebuch and Gall 1990) on the Fremont 
National Forest and adjacent private forest lands 
dating from 1969-92, we evaluated temporal 
changes to forest structure around goshawk nests 
during 1992-94. Our objectives were to: (1) deter- 
mine if a random sample of historical goshawk nest 
areas (i.e., nests first found in 1973-91) were oc- 
cupied in 1994, (2) document post-1992 forest con- 
ditions and quantify change in forest cover on his- 


torical nest areas, and (3) compare 1994 forest 
cover between historical nest areas that were oc- 
cupied by goshawks between 1992 and 1994 to 
those historical nest areas where presence of gos- 
hawks was not detected. 

Study Area 

Research took place on the Fremont National Forest 
(FNF) and the Klamath Tree Farm of the Weyerhaeuser 
Company in south-central Oregon, encompassing >5000 
km^. Elevations ranged from 1200-2200 m. Ponderosa 
pine {Pinus ponderosa) , white fir {Abies concolor) , and lod- 
gepole pine {P. contorta) were the dominant commercial 
tree species. Generally, large expanses of lodgepole pine 
interspersed with small stands of pure ponderosa pine 
on higher ground dominated the northern half of the 
study area; dry mixed-conifer stands (white fir, incense 
cedar [Libocedrus decurrens], ponderosa pine, and sugar 
pine [R lambertiana] ) dominated the southern half of the 
study area. Douglas-fir {Pseudotsuga menziesit) was rarely 
encountered or absent, and most adjacent private lands 
had extensive ponderosa pine plantations. Natural forest 
openings consisted of xeric rocky flats, which contained 
sagebrush {Artemisia spp.) and bitterbrush {Purshia triden- 
tata) near ponderosa pine and mixed-conifer stands, and 
moist meadows, which were typically associated with lod- 
gepole pine and had a vegetative cover of sedges ( Carex 
spp.), sagebrush, and willow {Salix spp.) next to peren- 
nial streams or springs. The landscape was a mosaic of 
forest cover types, containing two large burned areas 
from the 1950s and 1992, natural openings, and human- 
created openings. Dominant silvicultural practices on 
Forest Service lands were partial harvest, selective remov- 
al, and shelterwood treatments in mixed-conifer and 
ponderosa pine. All forest management terms used in 
this paper follow Helms (1998). Regeneration (clearcut) 
harvest was more typical in lodgepole pine habitat, al- 
though observational data and Forest Service records 
(Fremont National Forest Supervisor’s Office, Lakeview, 
OR U.S.A.) documented that regeneration harvest oc- 
curred in mixed conifer and mature ponderosa pine 
types. Private land management was dominated by mosdy 
early serai and some mid-seral plantations of P. ponderosa 


312 


Desimone and DeStefano 


VoL. 39, No. 3 


in large clearcuts or past overstory removal with few scat- 
tered large seed trees. Forest Service management, reg- 
ulated timber harvest and aggressive fire suppression 
dates back >50 yr; selective railroad logging took place 
around 1920 (Hopkins 1979, Laudenslayer et al. 1989). 
Regional historical accounts state that ponderosa pine 
stands were typically composed of large trees with a mean 
DBH of 40-70 cm and basal area (BA) ranging from 13 
to 23 m^/ha (Munger 1917), stands rarely encountered 
in managed forests during our study. 

Methods 

We defined nest site as the tree containing the occupied 
nest or the mapped location of the historically occupied 
nests and ^1 ha around the location. A nest area for this 
study was the area that we surveyed out to 1000 m (about 
300 ha), centered on a nest site. We defined post-1992 
nest areas as occupied (breeding) nest areas first discov- 
ered during our study, which was conducted during 
1992-94. Historical nest areas were defined as pre-1992 
nest areas if they were first discovered occupied 1973-91 
by Forest Service or Weyerhaeuser personnel, or other 
researchers. Occupied nest areas were areas we surveyed 
during 1992-94, where at least one adult goshawk was 
present or actively nesting. For purposes of habitat- 
change comparisons in 1994, occupied nest areas were a 
subset of the historical pre-1992 nest areas that were 
found occupied in 1994. No-response areas were a subset 
of pre-1992 nest areas surveyed in 1994 that had no de- 
tections. 

Goshawk Nest Area Occupancy Surveys. We compiled 
a list of historical goshawk nest area locations from orig- 
inal data collected by Reynolds (1975, 1978), U.S. Forest 
Service (unpublished data, Fremont National Forest, 
Lakeview, OR U.S.A.), and Weyerhaeuser Company (un- 
published data, Klamath Tree Farm, Klamath Falls, OR 
U.S. A.) and evaluated each dataset based on quality of 
documentation (e.g., written reports, legal and area de- 
scriptions, mapped locations), observer reliability (e.g,, 
biologist or experienced observer) , and number of years 
the nest area was documented as occupied. Nest records 
were included if there was adequate documentation of a 
goshawk attending a nest structure, incubating, or if 
fledglings or nestlings were present at the nest site. Lo- 
cations meeting the above criteria were mapped, and for- 
est cover type was validated by aerial photograph or field 
examination before surveys commenced. We stratified 
sites into one of three forest cover types: dry mixed-co- 
nifer, ponderosa pine (<20% other tree species), and 
lodgepole pine (<20% other tree species). 

We broadcast conspecific vocalizations to elicit respons- 
es from nesting goshawks or fledglings from late May to 
early August 1992-94 (Kennedy and Stahlecker 1993, Joy 
et al. 1994). Surveys were centered on the last known 
occupied historical nest location, with at least 35-40 call- 
ing stations per survey area (see below), spaced 320 m 
apart and staggered on adjacent and parallel transect 
belts. To ensure coverage of potential nest areas, we ex- 
amined the literature for estimates of inter-nest distances 
between alternative nest sites (273 m in the Klamath NF, 
California [Woodbridge and Detrich 1994]; 266 m [Reyn- 
olds et al. 1994], and 489 m [Reynolds and Joy 1998] in 
Arizona; and 432 m in Utah [Dewey et al. 2003]), sizes 


of post-fledging family areas (PFA; ca. 168 ha, Kennedy 
et al. 1994), and the effective auditory range of the mega- 
phone (^150 m; Joy et al. 1994; S. Desimone unpubl. 
data) . Based on this information, we established our sur- 
vey area size as a circle with an approximate 1000-m ra- 
dius centered around the nest location (ca. 40 stations). 
This resulted in a search area of about 300 ha, nearly 
twice the area of mean PFA size reported by Kennedy et 
al. (1994). If a response was detected, we immediately 
searched the vicinity for an occupied nest. For those sur- 
veyed areas where there were no detections during the 
nestiing period (first survey), we resurveyed the area at 
least once in July— August during the fledgling period us- 
ing the same stations so that each “no-response” area was 
visited and surveyed at least twice in a season. We also 
conducted systematic and opportunistic searches (De- 
Stefano et al. 1994a, Daw et al. 1998) for new goshawk 
nests (i.e., post-1992) during May-August 1992-94. When 
surveying a known occupied nest area from the previous 
survey season (i.e,, 1992 or 1993), we used multiple ob- 
servers to conduct a silent search at the last known oc- 
cupied site to minimize disturbance. If there were no de- 
tections, we extended the search pattern by radiating out 
from the nest tree while using a combination of inter- 
mittent taped broadcast calls near the nest and regularly 
spaced calling stations. These areas had the same level of 
survey effort as the systematic searches: about 300 ha 
around the last known occupied nest. 

Vegetation Sampling. We used 1:12000 and 1:15 800 
scale aerial photographs to describe and classify historical 
(1969-91) forest vegetation conditions and post-1992 
conditions, obtained from the U.S. Forest Service and 
Weyerhaeuser Company for years 1969, 1972, 1976, 1978, 
1980, 1983, 1988, and 1992 (the most recent available) 
for reference stands. Harvest inventory data from the 
Fremont National Forest were used to update 1988 and 
1992 photos to 1994 conditions. 

We used a 3X Dietzgen stereoscope to delineate cover 
in an 11% random sample (25 of 227) of reference stand 
polygons representing the range of forest conditions and 
habitats on the 1992 photographs. The variable-plot veg- 
etation sampling method (Bell and Dilworth 1988) was 
used to verify the condition of these reference polygons 
on the ground. We sampled 7-11 plots (x = 8.1 plots/ 
reference polygon, SD = 1.8), 160 m apart, on a transect 
located through the longest axis of the habitat polygon 
or in parallel transects if the polygon was >300 m wide. 
Plots were measured for basal area (BA) using a 20-factor 
(ft^/acre, later converted to m^/ha) wedge prism at plot 
center to sort trees into diameter classes. We recorded 
DBH for all count trees by combining plots within a stand 
to determine trees/ha (TPH) and BA for each forest 
structure class. The stem count per sample point multi- 
plied by the BA factor equaled the total BA occupied by 
tree stems on a per ha basis (Bell and Dilworth 1988). 
We followed the U.S. Forest Service Region 6 Vegetation 
Structural Stage (VSS) guidelines for general forest cover 
type descriptions in eastern Oregon (U.S. Department of 
Agriculture 1994). 

We used two non-forest categories (open wet [moist 
meadows] and open dry [xeric flats]) and four forest 
structure categories (late, mid-age, early, very-early), 
combined with two canopy closure classes (<50% or 


September 2005 


Conservation 


313 


Table 1. Forest structure classification for aerial photograph interpretation on the Fremont National Forest and 
adjacent private lands in Oregon U.S.A., based on mean tree diameter at breast height (DBH), mean canopy closure, 
and trees per ha (TPH) S;53 cm DBH (USDA 1994). Very early stage was forest regeneration or clearcut. 


Forest Structure 

DBH (cm) 

Canopy Closure 
(%) 

TPH >53 cm 

Late closed 

>53 

>50 

>15 

Late open 

>53 

<50 

>15 

Mid-aged closed 

23-53 

>50 

<15 

Mid-aged open 

23-53 

<50 

<15 

Early closed 

12-23 

>50 

none 

Early open 

12-23 

<50 

none 

Very early 

<12 

<50 

none 


>50%), to identify and delineate vegetative cover on ae- 
rial photographs. Stands were then typed into forest veg- 
etative cover classes based on total BA of trees per di- 
ameter class and TPH >53 cm (Table 1). We defined 
canopy closure as the amount of sky obscured by tree 
foliage and branches as measured by a Lemmon spheri- 
cal densitometer (Vales and Bunnell 1985). Canopy mea- 
surements were taken 5 m from plot center in four car- 
dinal directions, averaged, and mean percent canopy 
closure was calculated from all plots for the polygon. 

Following reference plot validation, all remaining hab- 
itat on photographs within a 170-ha circle around nest 
locations was delineated into vegetative cover polygons 
based on the validated reference plots and assigned veg- 
etation structure categories. When 1994 photographs 
were not available, the 1994 Fremont National Forest 
Harvest Inventory (U.S. Forest Service, Fremont NF, 
Lakeview, OR U.S.A.) was used to update the habitat con- 
dition. A 19% (AT — 102) random sample of polygons (N 
= 546), stratified by general forest cover type, was 
ground-verified using the same variable-plot sampling 
method outlined for reference stands. We assessed stand- 
typing accuracy by constructing an error matrix to deter- 
mine the accuracy of our photograph interpretation (De- 
simone 1997). 

To delineate historical forest conditions, we used U.S. 
Forest Service and Weyerhaeuser Company aerial pho- 
tographs (1:12000, 1:15 800, and 1:24000 scales) that 


represented stand conditions present in the year of the 
last known occupied nest. We extrapolated our reference 
set results to type stands into vegetative structure classes 
on the remaining historical photographs. All completed 
polygons were transferred to 1:24000 scale U.S. Geolog- 
ical Survey (USGS) quadrangle maps using a zoom trans- 
fer scope (Bausch and Lomb Corporation, Rochester, NY 
U.S.A.) and digitized into a Geographic Information Sys- 
tem, where area was calculated for each habitat polygon. 

Annual Variation in Occupancy. We estimated annual 
variation in occupancy by resident pairs of western North 
American goshawks by examining data from five other 
study areas (Table 2). We then compared the mean an- 
nual occupancy rates of goshawks from these five study 
areas to our findings for post-1992 nest areas and pre- 
1992 (historical) nest areas assessed in 1994. Annual oc- 
cupancy was defined as the mean (SE) annual percent 
of occupied areas. We assumed (1) territory occupancy 
was determined using similar survey techniques with 
equal effort (Joy et al. 1994, Reynolds et al. 1994, Wood- 
bridge and Detrich 1994, Kennedy 1997) and (2) littie 
or no major stand disturbance or habitat alteration oc- 
curred within territories since discovery (S. Dewey, R 
Kennedy, R. Reynolds, and B. Woodbridge pers. comm.). 
An occupied territory was defined by all researchers as 
>:! adult goshawk present in or near the nest on >2 
separate occasions during the breeding season and in- 


Table 2. Mean occupancy rates of Northern Goshawk nest areas (NA) among six concurrent studies in the western 
U.S., including occupied nest areas found on the Fremont National Forest (NF) and adjacent private lands 1992—94, 
Oregon, U.S.A. (this study). Occupancy is defined as a territory used regularly by at least one adult goshawk during 
the breeding season. The first year a nest was discovered is not included in the calculations. 


Study Area 

Source 

NA 

Mean 

SE 

Wars® 

Kaibab NF, AZ 

R.T. Reynolds pers. comm. 

32 

0.72 

0.05 

4-5 

New Mexico 

Kennedy 1997 

22 

0.74 

0.07 

4-11 

Klamath NF, CA 

Woodbridge and Detrich 1994 

26 

0.74 

0.01 

5-9 

Utah 

Kennedy 1997 

26 

0.75 

0.06 

4-7 

Malheur NF, OR 

S.K. Daw pers. comm. 

33 

0.66 

0.02 

2-4 

Fremont NF, OR 

This study 

20 

0.79 

0.04 

2 


^ Number of years of occupancy data for known nests in the study area. 


314 


Desimone and DeStefano 


VoL. 39, No. 3 


eluded pairs attempting to nest (Reynolds et al. 1994, 
Woodbridge and Detrich 1994, Kennedy 1997). 

Habitat Chaise Analysis. Using nest tree locations as 
nest area centers, we established five different radius cat- 
egories of 12, 24, 52, 120, and 170 ha. These areas had 
biological or managerial significance: 12 ha was recom- 
mended as a minimum nest area size for goshawks (Reyn- 
olds 1983, Reynolds et al. 1992); 24 ha was the size of 
goshawk habitat areas designated on the Fremont NF to 
protect nesting stands (U.S. Department of Agriculture 
[USDA] 1989); 52 ha was about the mean size of the ag- 
gregate of alternative nest areas associated with the pri- 
mary nest area of goshawks nesting in the Klamath NF in 
northern CA (Woodbridge and Detrich 1994); 120 ha was 
the area of old-growth habitat allocated for management 
of Pileated Woodpeckers (Dryocopus pileatus) on the Fre- 
mont NF (USDA 1989); and 170 ha was the size of the 
goshawk PEA (Reynolds et al. 1992, Kennedy et al. 1994). 

We conducted comparisons at both “disk” (12, 24, 52, 
120, and 170 ha) and “ring” (the area between the 12- 
24, 24-52, 52-120, and 120-170 ha disks) scales. Disks 
represent cumulative effects as scale increases, since 
smaller disks are included within the larger disks. Rings 
were tested individually so that influence of inner disks 
was removed (Ramsey et al. 1994, McGrath et al. 2003). 

We examined how forest structure around historical 
nest sites changed over time by calculating the % change 
for each vegetation cover (forest and non-forest) variable 
(equation 1): 

% Change = - Area^j^Tonic) / (1) 

AreafjisTORic^ X 100 

where is the area of a cover category for 1994, 

and AreaffisTORic is the area of the same cover category in 
the year the site was last known to be occupied. This 
calculation was made for each of the paired sites for all 
scales of disks and rings. We used Wilcoxon signed-rank 
test for paired comparisons to test for changes in forest 
cover between pre-1992 and post-1992 conditions and 
Kruskal-Wallis one-way analysis of variance by ranks fol- 
lowed by multiple comparison tests for least significant 
difference (LSD) to test for differences in forest cover 
among pre-1992, pre-1992 occupied, and pre-1992 no- 
response nest areas (Conover 1980; JMP Statistical Soft- 
ware version 3.1, SAS Institute, Cary, North Carolina, 
U.S.A). 

Logistic Regression Model. We wanted to know the 
likelihood of predicting the suitability of historical nest 
areas by considering the amount of area of each vegeta- 
tion structure category (forest and non-forest) around 
occupied and no-response sites. Thus, we constructed a 
logistic regression model (Hosmer and Lemeshow 1989, 
Ramsey et al. 1994) using the binary response variable of 
occupied (F = 1) or no-response (F = 0) by goshawks 
in a historical nest area in 1994. The importance of a 
particular habitat variable was determined by a stepwise 
analysis (PROC LOGISTIC; SAS Institute, Inc., 1992). 
The alpha for entry level (p^) of the variable to be con- 
sidered for the model was 0.15 because we wanted to 
detect possible trends in the event of a nonsignificant P- 
value. Models were run for each of the five disks and four 
rings. The full model included all explanatory habitat 
variables (equation 2); 


logit P (Y) = -I- X VeryEarly 

+ B 2 X EarlyClosed 

+ X EarlyOpen 

+ B4 X MidClosed 

+ B^ X MidOpen 

+ B^ X LateClosed 

-h By X LateOpen 

+ Bg X OpenWet 

+ B^ X OpenDry (2) 

where Bq is constant, and By through B^ are the coeffi- 
cients. The model was run in logit P(l) mode (stepwise 
descending) to calculate odds ratios for significant vari- 
able (s) associated with a nest area being occupied (i.e., 
F = 1). Interaction terms were evaluated in the final 
model. 

Results 

Nest Area Surveys. During 1992-94, we found 38 
occupied goshawk nest areas (15 pre-1992 and 23 
post-1992) that composed our sample of nests for 
vegetation analysis. Of the 51 pre-1992 nest areas 
we reexamined, 10 had evidence of nesting and 
five more had goshawks present, for a total of 15 
historical occupied nest areas. Twelve of the 15 oc- 
cupied nest areas were contained wholly on Forest 
Service ownership, two were on Weyerhaeuser 
land, and one was mixed ownership. Of 36 no-re- 
sponse areas, 23 were on Forest Service, 1 1 on Wey- 
erhaeuser, and two were mixed ownership. We re- 
moved five nests from the historical sample for our 
vegetation analysis because of inadequate photo- 
graphic records. Therefore, of the remaining 46 
pre-1992 nest areas surveyed to protocol, 15 were 
occupied, and 31 were no-response areas. Number 
of nestlings per nest was similar for pre-1992 areas 
occupied in 1994 and post-1992 nests (1.5 ± 1.2 
[N = 10] and 1.4 ± 1.0 [N = 18] young/nest, re- 
spectively) . 

Annual Variation in Territory Occupancy. Of 38 

occupied nest areas, we were able to consistently 
survey 20 for at least two seasons from 1992-94; 
these had a mean annual occupancy rate of 79% 
(SE = 4; Table 2). This was similar to the mean 
annual occupancy rates from five other concurrent 
studies in the western U.S. (73%, SE = 2, for Ari- 
zona, California, New Mexico, Utah, and Oregon; 
Table 2). Occupancy of all historical (pre-1992) 
nest areas surveyed in 1994 was 29% (15/51), 
which was significantly different from the occupan- 


September 2005 


Conservation 


315 


cy rate for post-1992 areas (x^ = 12.4, 1 df, P = 
0.0004) and substantially lower than reported in 
the literature (Table 2). 

For post-1992 nest areas, mean inter-alternative 
nest distance was 245 m (SE = 48, N = 23; no data 
for private lands). This was comparable to the in- 
ter-alternative nest distances reported in the liter- 
ature (Reynolds et al. 1994, Woodbridge and De- 
trich 1994, Dewey et al. 2003). 

Habitat Typing. Overall typing accuracy based 
on ground verification of reference polygons was 
80%. Mid-age and late categories were 80-90% for 
reference polygons (Desimone 1997); we thought 
this was an acceptable rate to proceed with the 
analysis (Lillesand and Kiefer 1994). Our highest 
classification accuracies were for dry open and wet 
open non-forest categories (100% each) from ref- 
erence polygons, followed by late open and late 
closed forest structure (90% each), mid-aged open 
(84%), and mid-aged closed forest (80%). Early 
open and early closed forest structure was least ac- 
curately classified(67% and 69%, respectively). 

Forest Cover Distribution. For post-1992 nest ar- 
eas, 25 of 42 (60%) occupied nest trees were within 
late closed forest structure, and 11 of 42 (26%) 
were in mid-aged closed structure. Distribution of 
post-1992 and pre-1992 nest sites was similar 
among the three forest cover types: 56% versus 
47% in mixed conifer, 24% versus 28% in lodge- 
pole pine, and 20% versus 25% in ponderosa pine, 
respectively. 

Habitat Change Analysis. Mean percent change 
of the seven forest-structure categories (Table 1) 
for pre-1992 nest areas over time occurred in all 
scales (i.e., five disks and four rings) (see Desi- 
mone 1997: Tables 11 and 12 for details). For disks, 
the largest increases were in the amounts for very- 
early (642%, SE = 93%) and early open (238%, 
SE = 17%) categories. The largest decreases over 
time were in the late open (—54%, SE = 3%), late 
closed (—49%, SE = 1%), and mid closed ( — 30%, 
SE = 3%) categories. The magnitude of the per- 
cent change decreased with increasing scale; for 
example, increase in very-early cover went from 
742% to 640% to 435% at 12-, 52-, and l70-ha disk 
scales, respectively, while decreases in late open 
cover went from —58% to —56% to —47% for 12-, 
52- and 170-ha scales, respectively. Similar results 
were noted for rings, although at lower magni- 
tudes. 

In Figure 1, we presented late closed, early open, 
and very-early structural stages because they rep- 


40 
35 ■; 


Late Closed 


30 i 
25 ■{ 
20 -I 


$ 


15 ■ 



♦ Pre-'i99.' 

□ Occupied 
A No response 


♦ 

0 

A. 

I 


? 

T 

I 


•f 



t 





12 24 52 120 170 

Scale (ha) 


Figure 1, Mean (SE) area of late closed, early open, and 
very early structural stages among 5 circular analysis 
scales surrounding 46 historical (pre-1992; first discov- 
ered during 1973-91) goshawk nest sites in south-central 
Oregon, U.S.A.; 15 were occupied by goshawks and 31 
had no evidence of occupancy in 1994. See text and Ta- 
ble 1 for further description of forest structure catego- 
ries. 


resented most confidence in correctly classifying 
habitat types, and therefore most confidence in de- 
tecting a decrease in area of highest suitable hab- 
itat (late closed) and an increase in area of known 
non-nesting habitat (early open and very early). 
For the 12-, 24- and 52-ha scales, mean percent late 
closed forest at all occupied nest areas in 1994 re- 
mained nearly the same as at pre-1992 areas (i.e., 
no significant difference) . However, mean percent 
late closed forest at no-response areas was about 


316 


Desimone and DeStefano 


VoL. 39, No. 3 


80 ■ 
70 ■ 
60 
50 
40 .! 
30 .| 
20 ! 


12h* 


• k 


d 

k. 

u 


I 


10 i 


0 

60 
50 • 
40 
30 ■ 
20 . 
10 ■ 
0 • 



60 ■ 



Very Early Early Open Mid-age Late Open Late Closed Mid Closed 
Closed + Late 

Closed 


Forest Structural Stage 

Figure 2. Mean (SE) distribution of forest structural 
stage categories (plus a combination of mid-aged and late 
closed canopy forest) at 12-, 52 -, and l70-ha scales sur- 
rounding 46 historical (pre-1992 — first discovered during 
1973—91) goshawk nest sites in south-central Oregon, 
U.S.A.; 15 were occupied by goshawks and 31 had no 
evidence of occupancy in 1994. We omitted the 52- and 
120-ha data because results were similar. See text and Ta- 
ble 1 for further description of forest structure catego- 
ries. Difference in grouped means assessed by Kruskal- 
Wallis test (a = 0.05) . Within each group, Fisher’s test of 
least significant difference for multiple comparisons was 
used; pairs within each forest structure stage not signifi- 
candy different share common letters. 


one-fourth to one-fifth the amount of late closed 
at pre-1992 and occupied pre-1992 areas (Fig. 1). 
With increasing scale, the mean proportion of ear- 
ly open structure at no-response nest areas was 4- 
5 times greater than pre-1992 nest areas, and over 
twice that of occupied pre-1992 areas (Fig. 1). The 
mean proportion of very early stage increased with 
increasing scale and was 4-6 times greater at 12-, 
24- and 52-ha scales for no-response areas than pre- 
1992 and occupied pre-1992 areas (Fig. 1). For 
120- and 170-ha scales, mean proportion of very 
early was about half that in occupied compared to 


no-response areas; pre-1992 areas had about 1/8*^ 
that of no-response areas. 

No-response areas (N = 31) showed significant 
changes in the general distribution of forest struc- 
ture compared to all pre-1992 areas and also dif- 
fered significantly from occupied pre-1992 areas 
for 12-, 52- and 1 70-ha disk scales (Kruskal-Wallis 
test, all P < 0.0454; Fig. 2). Histograms for 24- and 
120-ha scales were not presented, as patterns in re- 
sults were similar but intermediate in values be- 
tween their adjacent scales. Mean proportion of 
area was significantly different for late closed for- 
est, mid-aged closed forest, early open forest, and 
very early among pre-1992, occupied pre-1992, and 
no-response areas (LSD for pairs of means, P < 
0.05) . The greatest single change in a category was 
in the mean amount of late closed forest in pre- 
1992 (x = 22-27% among all disk scales) and no- 
response areas (x = 6—8%, all disks; Kruskal-Wallis, 
P< 0.0003). 

At the 52-ha scale, the mean percent area of late 
closed forest (20%) for occupied pre-1992 areas 
remained similar to historical pre-1992 areas (24%; 
Fig. 2). Less than half of the mean area of mid- 
aged closed forest that once existed in pre-1992 
areas (25%) occurred in no-response areas (12%). 
This corresponded with an increase in mean per- 
cent mid-aged open forest in no-response (24%) 
compared to the occupied pre-1992 areas (15%). 
In no-response areas, mean percent area of early 
open canopy forest was >4 times the historical 
mean (pre-1992) amount, and more than twice 
that of occupied pre-1992 areas (LSD test of 
means, P < 0.05). Very early mean percent area 
was significantly greater in no-response than oc- 
cupied pre-1992 areas (LSD test of means, P < 
0.05; Fig. 2). We point out comparisons at the 52- 
ha scale because it represents, in theory, the ag- 
gregate of alternative nest sites for a nesting area, 
and the persistence of goshawk use or occupancy 
appears to be correlated with higher amounts of 
mature forest at about this scale (Woodbridge and 
Detrich 1994). 

Logistic Regression Model of Forest Structure 
Association. For post-1992 occupied nest areas, 
both late and mid-aged closed variables were as- 
sociated with the 52-ha disk model (drop in Devi- 
ance = 9-5; 1 df; P < 0.01) and the 24—52 ha 
ring model (drop in Deviance = 20.7; 1 df; P < 
0.01), as described by the reduced model (Equa- 
tion 3): 



September 2005 


Conservation 


317 


Table 3. Results of stepwise logistic regression analysis for occupied (F = 1) goshawk nest areas (A^ = 15), Fremont 
National Forest and adjacent private lands, Oregon, U.S.A., 1994. Stepwise entry level was at a = 0.15. Scales emanate 
from territory centers; ring size is the area between two concentric disk areas. Parameter estimates are natural log 
(In) of odds ratios. The interaction term (late closed X mid-aged closed) was not significant (P = 0.23). 


Disk Size® 
(ha) 

Variable 

Parameter 

Estimate 

SE 

Wald 

P-value 

Odds 

Ratio 

Estimate 

95% 

Confidence 
Interval, of 
Odds Ratio 

12 

Intercept 

-83.9333 

24.4576 

11.7771 

0.0006 

— 

— 


Late 

closed 

0.4771 

0.1650 

8.3594 

0.0038 

1.611 

0.1537-0.8005 

(1.166-2.227) 


Mid-aged 

closed 

0.3344 

0.1157 

8.3554 

0.0038 

1.397 

0.1076-0.5612 

(1.114-1.753) 

24 

Intercept 

Late 

-46.5816 

13.664 

11.6205 

0.0007 

— 

0.0804-0.4516 


closed 

Mid-aged 

0.2660 

0.0947 

7.8850 

0.005 

1.305 

(1.084-1.571) 

0.0522-0.2936 


closed 

0.1729 

0.0616 

7.8829 

0.005 

1.189 

(1.054-1.341) 

52 

Intercept 

-21.9700 

6.3879 

11.8290 

0.0006 

— 

— 


Late 

closed 

0.1131 

0.0401 

7.9426 

0.0048 

1.120 

0.0345-0.1917 

(1.035-1.211) 

Ring Size^ 
(ha) 

Mid-aged 

closed 

0.0818 

0.0307 

7.1046 

0.0077 

1.085 

0.0216-0.1420 

(1.022-1.155) 

12-24 

Intercept 

-85.3932 

25.1893 

11.4925 

0.0007 

— 

— 


Late 

closed 

0.5126 

0.1798 

8.1303 

0.0044 

1.670 

0.2109-0.9366 

(1.235-2.551) 


Mid-aged 

closed 

0.3175 

0.1215 

6.8264 

0.009 

1.374 

0.0984-0.5913 

(1.103-1.806) 

24-52 

Intercept 

-33.9116 

10.7673 

9.9193 

0.0016 

— 

— 


Late 

closed 

0.1754 

0.0691 

6.4489 

0.0111 

1.192 

0.0533-0.3301 

(1.055-1.391) 


Mid-aged 

closed 

0.1423 

0.0579 

6.0437 

0.0140 

1.153 

0.0376-0.2719 

(1.038-1.313) 


■^Forest structure classes were not significantly associated with the 120- and l70-ha disk scales. 
’’Forest structure classes were not significantly associated with the 52-120 and 120-170 ring scales. 


logit (1) — Bq + (late closed) 

+ B 2 (mid-aged closed) 

There was a strong association between nest area 
occupancy and both late closed and mid-aged 
closed forest at the 12, 24, and 52 ha scales (Table 
3) . At the 12-ha nest area scale, the odds that a site 
was occupied increased by 61% (odds ratio 1.61) 
for each unit ( 1 ha) increase of late closed forest 
habitat, while holding the mid-aged closed forest 
variable constant. For each unit increase of mid- 
aged closed forest habitat, the odds that a site was 
occupied increased by 37% (odds ratio 1.37), while 
holding the late closed forest variable constant. 


The reduced model was also significant for the 
12—24 and 24—52 ha rings. The stepwise descend- 
ing model procedure did not yield a significant 
model for any variables associated with occupied 
sites for 120 or 170 ha disks, or for 52-120 and 
120-170 ha rings. The interaction term {late closed 
X mid-aged closed) of the reduced model for disks 
was not significant (x^ = 43.1; 1 df; P = 0.23). 

Discussion 

Not all goshawk territories may be occupied in 
all years (Detrich and Woodbridge 1994, Reynolds 
et al. 1994), and even in the absence of human- 
caused habitat alteration, some territories can be 


318 


Desimone and DeStefano 


VoL. 39, No. 3 


expected to be lost due to natural (e.g., stand se- 
nescence, disease, fire) changes in the forest over 
time (Graham et al. 1994). Mean annual occupan- 
cy of goshawk nesting areas in six studies across the 
western U.S. (concurrent with this study) were con- 
sistently in the 65-80% range over 2-11 yr of study. 
An occupancy rate of 29% (15/51) of the historical 
nest areas on our study area in 1994 is low by com- 
parison. The low occupancy rate may be due in 
part to attrition of some suitable nest areas due to 
natural disturbance over time; one fire partially 
burned two nest areas in our study. It is also pos- 
sible that our goshawk surveys did not extend out 
far enough to include some of the alternative nests 
used. Recent data from Arizona suggests that about 
67% of goshawks move to alternative nest locations 
every year and that a 1000 m broadcast calling ra- 
dius accounted for about 95% of the alternative 
nest attempts (R. Reynolds unpubl. data). If these 
findings are applied to our study, we likely missed 
about 5% of all alternative nests in our study. 

Like many raptors, individual goshawk pairs may 
not breed every year, and determining trends in 
territory occupancy using 2 yr survey data is ten- 
tative (DeStefano et al. 1994a). Pairs not nesting 
in a given year, but still occupying the nest area, 
are difficult to hnd when surveys are conducted 
after courtship (Dewey et al. 2003). However, we 
searched large areas (>300 ha) multiple times 
around each historical nest location during a pe- 
riod when local weather conditions were not par- 
ticularly inclement for the region and when gos- 
hawk productivity was relatively high: 74% (17/23) 
of nests on the Fremont NF and 91% (20/22) of 
nests on the nearby Malheur National Forest suc- 
cessfully fledged young in 1994 (S. Rickabaugh, S. 
Danver, and S. Daw unpubl. data). Other studies 
m eastern Oregon and Washington reported simi- 
lar high occupancy and nest success levels for 1994 
(McGrath et al. 2003; S. Finn unpubl. data) . In ad- 
dition, Kostrzewa and Kostrzewa (1990) reported 
that weather did not affect the density of territorial 
goshawk pairs over an 8-yr period in Europe, but 
was an influential limiting factor to breeding suc- 
cess. Thus, we concluded that the low occupancy 
rates of the historical nest areas were not attributed 
to low detectability, although we could not com- 
pletely rule this out as a possibility. 

The difference in forest structure between post- 
1992 occupied and no-response nest areas was 
compelling. Late structural stage forest, especially 
with canopy cover >50%, was much more preva- 


lent around occupied than no-response nest areas. 
Conversely, very early and early structural stage for- 
est was much more prevalent in no-response than 
occupied nest areas. Our results indicated that late 
forest structure declined by 20-50%, and very early 
and early forest structure increased by >400% 
around no-response nests. These trends were de- 
tectable at all scales, but were strongest at the 
smaller scales (12 and 52 ha) and decreased with 
increasing scale. Although we do not have detailed 
history of stand management for all cases, the ob- 
served difference in habitat is attributed to levels 
of timber harvest, which we verified by photo- 
graphic evidence and field examination. The loss 
of large trees (>53 cm DBH) and a reduction in 
canopy cover to <50% appeared to influence nest 
area occupancy. Penteriani and Faivre (2001) and 
Penteriani et al. (2002) found that nest sites (ca. 
0.8 ha) around the nest tree altered by more than 
30%, either by selective tree harvest (shelter wood) 
or windthrow, caused goshawk pairs to change lo- 
cations to new nest stands. The general conclu- 
sions reached by Penteriani and Faivre (2001) and 
Penteriani et al. (2002) on habitat disturbance 
were consistent with our results: goshawks were ab- 
sent from nest areas where there was ^30% mean 
decrease in late and mid closed forest (12-ha scale) 
compared to the pre-1992 condition. Our data 
showed that this pattern was consistent at larger 
(12-52 ha) scales as well. 

Our results suggested that nest area habitat al- 
teration (loss of nesting habitat) was the most likely 
reason for the low occupancy rates of historical 
nest areas in 1994. The habitat alteration was likely 
the result of timber harvest (documented by aerial 
photographs), which reduced the proportion of 
late and mid-aged forest with high canopy closure 
and increased the proportion of very early and ear- 
ly open forest conditions within 52 ha (scale of lo- 
gistic model significance) of goshawk nests. 

Management practices for nesting habitat pro- 
tection on the Fremont NF were limited during 
1973-91, ranging from no protection (unrestricted 
harvest) of nest areas to 12-ha no-harvest buffers 
around nests during the breeding season (Reyn- 
olds 1983, USDA 1993). In 1983, the Fremont 
National Forest Plan established several 24-ha gos- 
hawk habitat management areas. However, condi- 
tions on most of these management areas ranged 
from early successional forests (unsuitable to mar- 
ginal for nesting habitat) to mid-aged forest with 
only small patches of late-successional forest. Some 


September 2005 


Conservation 


319 


goshawk management areas were reassigned or re- 
located in subsequent years to achieve timber har- 
vest objectives (K. Palermo and W. Watkins, Fre- 
mont NF, pers. comm., S. Desimone unpubl. data). 

The photographic record revealed that little or 
no long-term habitat protection was implemented 
for the 31 no-response areas as of 1994. All were 
historical sites that had some portions within 52 ha 
of the nest site harvested during or after the his- 
torical nesting season. In contrast, most goshawk 
territories in the western U.S. study areas we re- 
viewed (Table 2) had little or no habitat loss from 
timber harvest practices since discovery by the re- 
searchers and had yearly monitoring programs that 
documented relatively high occupancy rates (B. 
Woodbridge, P. Kennedy, R. Reynolds, and S. Dew- 
ey pers. comm.). This further supported our con- 
clusion that timber harvest was a determining fac- 
tor leading to significantly lower occupancy rates 
in the no-response nest areas compared to the oc- 
cupied areas. 

Nest area fidelity (as indexed by occupancy 
rates) is likely to be higher in good quality habitats 
as compared to poor quality habitats. This may be 
advantageous because there is an increased likeli- 
hood of nesting success where they may have been 
successful before (Newton 1979, Newton and Wyl- 
lie 1992, Rosenfield and Bielefeldt 1996). Our re- 
sults suggest nest areas with >50% proportion of 
older and larger structural classes may be higher 
quality nest areas than areas dominated by younger 
serai stages (Woodbridge and Detrich 1994, Finn 
et al. 2002). Detrich and Woodbridge (1994) and 
Reynolds et al. (1994) reported that 70-75% of 
banded goshawks occupied the same nest area in 
successive years, which was similar to findings for 
Cooper’s Hawks (A. cooperii; Rosenfield and Biele- 
feldt 1996) and Eurasian Sparrowhawks (A. nisus\ 
Newton and Wyllie 1992). Although anecdotal, in 
1992-94 we found an occupied nest in each of two 
nest areas that were both within 100 m of their 
respective historical nest site in nest areas that re- 
ceived special protection as old growth manage- 
ment areas in the early 1980s (Fremont NF un- 
publ. data) ; these sites were first found 20 yr earlier 
by Reynolds (1975). 

In Arizona, Reynolds and Joy (1998) reported 
that over a 6-yr period, 92% of breeding male and 
79% of breeding female goshawks had fidelity to 
their territories and mates. However, in extreme 
conditions such as food stress (Newton 1979) or in 
disturbed habitats (Woodbridge et al. 1988, Bosa- 


kowski et al. 1993, Woodbridge and Detrich 1994, 
Crocker-Bedford 1995), there is evidence to sug- 
gest that species with strong site fidelity might be- 
have differently. Bosakowski et al. (1993) reported 
five of six Cooper’s Hawk nest sites were aban- 
doned and not reused in the year following clear- 
ing of adjacent forests and human encroachment 
within a range of 40-500 m of the nest site. Hargis 
et al. (1994) postulated that monitoring site fidelity 
of breeding goshawks might provide a valuable in- 
dicator of the quality of the surrounding home 
range. If specific habitats needed for foraging and 
development of fledglings are subjected to habitat 
alteration outside nest areas (defined as >12 ha m 
Hargis et al. [1994]), hawk pairs might vacate even 
though individual nest sites (i.e., <12 ha) are be- 
ing protected (Woodbridge et al. 1988, Bosakowski 
et al. 1993, Hargis et al. 1994, Woodbridge and De- 
trich 1994, Crocker-Bedford 1998). 

To infer that goshawk populations have declined 
on our study area is beyond the scope of this study. 
It is possible that goshawks not found in our his- 
torical no-response nest areas in 1994 had relocat- 
ed to more suitable areas elsewhere. However, in 
these no-response nest areas, forest structural con- 
ditions were significantly altered from past timber 
harvest, suggesting that habitat quality had been 
substantially reduced, which precluded goshawks 
from occupying those nest areas (i.e., out to the 
300-ha surveyed area in our study) through time. 

Our results indicated that pre-1992 nest areas 
still occupied by goshawks in 1994 had >50% of 
their mean area in mid closed + late closed forest 
within the 52-ha scale (Fig. 2), and most resembled 
their historical photograph conditions. Moreover, 
late forest (i.e., late closed and late open) structure 
was most predominant in occupied nesting areas 
at the 12-ha scale for all forest cover types exam- 
ined, supporting studies in Oregon (Moore and 
Henny 1983, Bull and Hohmann 1994, Daw and 
DeStefano 2001), northern California (Wood- 
bridge and Detrich 1994), and elsewhere (Reyn- 
olds et al. 1982, 1994, Crocker-Bedford and Cha- 
ney 1988, Hayward and Escano 1989, Siders and 
Kennedy 1996, Squires and Ruggiero 1996). In 
1994, only 2—8% of the forested area in the Ere- 
mont NE was composed of ponderosa pine or pine- 
associated, late structured, old forest (Henjum et 
al. 1994). Because of the decline of areas of con- 
tiguous large and old trees (>50 cm DBH or >150 
yr of age; Henjum et al. 1994), late-successional 
and old ponderosa pine forest has become an in- 


320 


Desimone and DeStefano 


VoL. 39, No. 3 


creasingly threatened forest ecosystem in North 
America (Noss et al. 1995). 

Implications of Vegetative Cover Loss. A mosaic 
of vegetative cover best describes goshawk nest ar- 
eas (i.e., 170 ha) on the Fremont and private lands 
we examined. For a goshawk population to persist 
in this mosaic, sufficient breeding habitat must ex- 
ist to promote positive net reproduction (Rosen- 
zweig 1985, Urban and Shugart 1986). Although 
recent analyses of goshawk demography in the U.S. 
reported no evidence of population trends (De- 
Stefano et al. 1994b, Kennedy 1997), forest man- 
agement activities such as intensive harvest and 
road building, as well as human development in 
the last 50-100 yr have changed the forest mosaic 
proportions to a far greater degree than natural 
disturbance regimes. In recent decades, for exam- 
ple, older forest has been harvested at a more rap- 
id rate than it can develop (USDA 1993, Henjum 
et al. 1994, DellaSalla et al. 1995). The accelerated 
pace of habitat change has greatly increased the 
proportion of early successional forest and resulted 
in a skewed distribution favoring younger age clas- 
ses compared to what was present historically in 
our study area (Henjum et al. 1994, 1996, Noss et 
al. 1995, Weyerhaeuser Corporate Photographic 
Archives unpubl. data). The net effect is that suit- 
able nesting and foraging habitat for goshawks is 
reduced (McCarthy et al. 1989, DellaSalla et al. 
1995, Henjum et al. 1996), and positive net repro- 
duction of goshawks and other species that use old- 
er forests is potentially affected. 

Our results lend evidence to the hypothesis that 
long-term occupancy of nest areas is correlated 
with larger proportions of mature forest (Wood- 
bridge and Detrich 1994) and indicates that sub- 
stantial amounts of late and mid-aged closed forest 
were important to the continued use of historical 
nest areas by goshawks. Significant differences in 
the amounts of mid-age closed and late closed for- 
est between historical (pre-1992) and occupied 
(post-1992) nest areas were not apparent in 1994 
at the 52-ha scale (Fig. 1, 2), suggesting that rela- 
tively intact forest structure resembling historical 
conditions contributes to its persistent use by gos- 
hawks. However, there was a slight significant dif- 
ference at the 170-ha scale for late closed and mid 
closed + late closed habitat. We could not predict 
the response of goshawks to limited alterations of 
habitat (e.g., thinning, light selection harvest). 
However, tree harvest prescriptions that create 
large areas with sparse cover are potentially detri- 


mental to nest area occupancy in our study area, 
especially if the percent of open canopy forest (i.e., 
very early, early open, mid open, late open) is 
>34% (mean) of the 52-ha scale or >44% (mean) 
of the 170-ha scale (Fig. 2). 

Management Implications. Our results showed 
that the presence of late and mid-aged closed for- 
est (combined, 60% and 48% within the 12-ha and 
52-ha scale, respectively) were important to the 
continued use of historical nest areas by goshawks. 
We recommend a no-harvest zone within the 12-ha 
around nest sites and discourage further cutting of 
large trees within the 52 ha. These recommenda- 
tions would help to preserve stand integrity, main- 
tain closed canopies, promote connectivity to al- 
ternative nest stands, and maximize conditions for 
breeding goshawk pairs to persist. Retaining exist- 
ing mid-aged closed and late closed forest struc- 
ture to levels of >50% at the 52-ha scale and >40% 
within the 170-ha scale, as well as managing to pro- 
mote this structure in the future, would also likely 
beneht goshawks. Based on our results, we also rec- 
ommend that about 10-20% of the surrounding 
forest structure outside the nest site be in very ear- 
ly or early open categories with the lesser amounts 
in the smaller scales (12 and 24 ha; Fig. 1). Man- 
agement within the 170-ha scale should be limited 
to light thinning or carefully prescribed burning 
of overstocked stands outside of the breeding sea- 
son (October-February) to promote mature, un- 
even-aged stand development. This could also im- 
prove foraging opportunities for goshawks by 
removing some of the dense understory of shade 
tolerant conifers. 

Finally, logistic regression analyses suggest that 
habitat alteration that reduces the proportion of 
mature closed-canopy forest, and which is subse- 
quently replaced by early successional forest, re- 
duces the probability of an area as a potential nest- 
ing habitat for breeding goshawks, supporting 
McCarthy et al. (1989). More severe alterations 
(clearcuts and moderately high alteration, partial 
removal of stands resulting in <50% canopy clo- 
sure) increase the likelihood of goshawks not re- 
occupying areas due to deterioration in the quality 
of potential nest-areas. 

Acknowledgments 

Funding was provided by the Oregon Department of 
Fish and Wildlife, U.S. Fish and Wildlife Service, Oregon 
State University Center for Analysis of Environmental 
Change, U.S. Forest Service Pacific Northwest Region, 
and the U.S. Geological Survey Biological Resources Di- 


September 2005 


Conservation 


321 


vision. We are indebted to E. Beck, C. Carstarphen, J. 
Citta, S.K. Daw, A. Krawarik, J. Mauer, M.T. McGrath, 
M.A. McLeod, C.D. Schelz, and K.H. Schmidt for field- 
work and data collection and the personnel and staff of 
the U.S. Forest Service and Weyerhaeuser Company for 
providing information on goshawks, land access, and ae- 
rial photographs. T. Hershey, P. Cooler, and D. Rau of 
the Fremont National Forest provided logistical support, 
the GIS data summary and final habitat maps. We thank 
R.J. Steidl for assistance with data analyses and reviewers 
D.A. Boyce, P.L. Kennedy, and D.E. Varland who helped 
improve the manuscript. 

Literature Cited 

Bell, J.F. and J.R. Dilworth. 1988. Log scaling and tim- 
ber cruising. Oregon State Univ. Bookstores, Inc., 
Corvallis, OR U.S.A. 

Bosakowski, T., R. Speiser, D.G. Smith, and L.J. Niles. 
1993. Loss of Cooper’s Hawk nesting habitat to sub- 
urban development: inadequate protection for a state- 
endangered species./. Raptor Res. 27:26-30. 

Bull, E.L. andJ.H. Hohmann. 1994. Breeding biology of 
Northern Goshawks in northeastern Oregon. Stud. 
Avian Biol. 16:103—105. 

Conover, W.J. 1980. Practical nonparametric statistics. 

John Wiley and Sons, New York, NY U.S.A. 
Crocker-Bedford, D.C. 1990. Goshawk reproduction 
and forest management. Wildl. Soc. Bull. 18:262-269. 

. 1995. Northern Goshawk reproduction relative 

to selection harvest in Arizona. /. Raptor Res. 29:42— 
43. 

. 1998. The value of demographic and habitat 

studies in determining the status of Northern Gos- 
hawks {Accipiter gentilis atricapillus) with special refer- 
ence to Crocker-Bedford (1990) and Kennedy (1997). 
J. Raptor Res. 32:329-336. 

AND B. Chaney. 1988. Characteristics of goshawk 

nesting stands. Pages 210-217 in R.L. Glinski, B.G. 
Pendleton, M.B. Moss, M.N. LeFranc, Jr., B.A. Millsap, 
and S.W. Hoffman [Eds.] , Southwest raptor manage- 
ment symposium and workshop. National Wildlife 
Federation, Scientific and Technical Series No. 11, 
Washington, DC U.S.A. 

Daw, S.K. AND S. DeStefano. 2001. Forest characteristics 
of Northern Goshawk nest stands and post-fledging 
areas in Oregon./. Wildl. Manag. 65:59—65. 

, , and RJ. Steidl. 1998. Does survey meth- 
od bias the description of Northern Goshawk nest-site 
structure? /. Wildl. Manag. 62:1379-1384. 

DellaSalla, D.A., D.M. Olson, S.E. Barth, S.L. Crane, 
AND S.A. Primm. 1995. Forest health: moving beyond 
the rhetoric to restore healthy landscapes in the in- 
land Northwest. Wildl. Soc. Bull. 23:346-356. 
Desimone, S.M. 1997. Occupancy rates and habitat rela- 
tionships of Northern Goshawks in historic nesting 
areas in Oregon. M.S. thesis, Oregon State Univ., Cor- 
vallis, OR U.S.A. 

DeStefano, S. 1998. Determining the status of Northern 


Goshawks in the west: is our conceptual model cor- 
rect? /. Raptor Res. 32:342—348. 

, S.K. Daw, S.M. Desimone, and E.C. Meslow. 

1994a. Density and productivity of Northern Gos- 
hawks: implications for monitoring and management. 
Stud. Avian Biol. 16:88-91. 

, B. Woodbridge, and P.J. Detrich. 1994b. Surviv- 
al of Northern Goshawks in the southern Cascades of 
California. Stud. Avian Biol. 16:133—136. 

Detrich, P.J. and B. Woodbridge. 1994. Territory fidel- 
ity, mate fidelity, and movements of color-marked 
Northern Goshawks in the southern-central Cascades 
of California. Stud. Avian Biol. 16:130—132. 

Dewey, S.R., P.L. Kennedy, and R.M. Stephens. 2003. Are 
dawn vocalization surveys effective for monitoring 
goshawk nest-area occupancy? /. Wildl. Manag. 67: 
390-397. 

Finn, S.P., J.M. Marzluff, and D.E. Varland. 2002. Ef- 
fects of landscape and local habitat attributes on 
Northern Goshawk site occupancy in western Wash- 
ington. Forest Science 48:427—436. 

Graham, R.T., R.T. Reynolds, M.H. Reiser, R.L. Bassett, 
AND D.A. Boyce. 1994. Sustaining forest habitat for 
the Northern Goshawk: a question of scale. Stud. Avi- 
an Biol. 16:12-17. 

Hargis, C.D., R.D. Perloff, and C. MgGarthy. 1994. 
Home ranges and habitats of Northern Goshawks in 
eastern California. Stud. Avian Biol. 16:66—74. 

Hayward, G.D. and R.E. Escano. 1989. Goshawk nest-site 
characteristics in western Montana and northern Ida- 
ho. Corzcior 91:476— 479. 

Helms, J.A. (Ed.). 1998. The dictionary of forestry. So- 
ciety of American Foresters, Bethesda, MD U.S. A. 

Henjum, M.G., J.R. Karr, D.L. Bottom, D.A. Perry, J.C. 
Bednarz, S.G. Wright, S.A. Beckwitt, and E. Beck- 
WITT. 1994. Interim protection for late-successional 
forests, fisheries, and watersheds: national forests east 
of the Cascade crest, Oregon and Washington. The 
Wildlife Society, Bethesda, MD U.S.A. 

, D.A. Perry, J.R. Karr, D.L. Bottom, J.C. Bed- 
narz, S.G. Wright, S.A. Beckwitt, and E. Beckwitt. 
1996. Maintaining ecological integrity of inland forest 
ecosystems in Oregon and Washington. Wildl. Soc 
Bull. 24:227-232. 

Hopkins, W.E. 1979. Plant associations of the Fremont 
National Forest. USDA Forest Service, Washington, 
DC U.S.A. 

Hosmer, D.W. and S. Lemeshow. 1989. Applied logistic 
regression. John Wiley and Sons, New York, NYU.S.A 

Joy, S.M., R.T. Reynolds, and D.G. Leslie. 1994. North- 
ern Goshawk broadcast surveys: hawk response vari- 
ables and survey cost. Stud. Avian Biol. 16:24-30. 

Kennedy, P.L. 1997. The Northern Goshawk {Accipiter gen- 
tilis atricapillus) : is there evidence of a population de- 
cline? /. Raptor Res. 31:95—106. 

and D.W. Stahlecker. 1993. Responsiveness of 


322 


Desimone and DeStefano 


VoL. 39, No. 3 


nesting Northern Goshawks to taped broadcasts of 
three conspecific calls./. Wildl. Manag. 57:249—257. 

, J.M. Ward, G.A. Rinker, and J.A. Gessaman. 

1994. Post-fledging areas in Northern Goshawk home 
ranges. Stud. Avian Biol. 16:75-82. 

Kostrzewa, a. and R. Kostrzewa. 1990. The relationship 
of spring and summer weather with density and 
breeding performance of the buzzard Buteo buteo, gos- 
hawk Accipiter gentilis, and kestrel Falco tinnunculus. Ibis 
132:550-559. 

Laudenslayer, W.F., Jr., H.H. Darr, and S. Smith. 1989. 
Historical effects of forest management practices on 
eastside pine communities in northeastern California. 
Pages 26-34 in A. Tecle, W.W. Covington, and R.H. 
Hamre [Tech. Coords.] , Multiresource management 
of ponderosa pine forests. USDA Forest Service, Gen- 
eral Technical Report RM-185, Rocky Mtn. Forest and 
Range Experiment Station, Fort Collins, CO U.S.A. 

Lillesand, T.M. and R.W. Kiefer. 1994. Remote sensing 
and image interpretation. Wiley and Sons, New York, 
NY U.S.A. 

McCarthy, C., W.D. Carrier, and W.F. Laudenslayer, Jr. 
1989. Coordinating timber management activities 
with raptor nesting habitat requirements. Pages 229- 
235 in B.G. Pendleton [Ed.] , Proceedings of the West- 
ern Raptor Management Symposium and Workshop, 
National Wildlife Federation Scientific and Technical 
Series No. 12, Washington, DC U.S.A. 

McGrath, M.T., S. DeStefano, R.A. Riggs, L.L. Irwin, 
AND G.J. Roloff. 2003. Spatially explicit influences on 
Northern Goshawk nesting habitat in the interior Pa- 
cific Northwest. Wildl. Monogr. 154:1-63. 

Moore, K.E. and CJ. Henny. 1983. Nest-site character- 
istics of three coexisting accipiter hawks in northeast- 
ern Oregon./. Raptor Res. 17:65-76. 

Monger, XT. 1917. Western yellow pine in Oregon. U.S. 
Department of Agriculture, Bulletin 418, Washington, 
D.C. U.S.A. 

Newton, I. 1979. Population ecology of raptors. Buteo 
Books, Vermillion, SD U.S.A. 

AND W.E. Wyllie. 1992. Fidelity to nesting terri- 
tory among European sparrowhawks in three areas./. 
Raptor Res. 26:108-114. 

Noss, R., E.T. LaRoe, III, and J.M. Scott. 1995. Endan- 
gered ecosystems of the United States: a preliminary 
assessment of loss and degradation. Biological Report 
28, National Biological Service, U.S. Department of 
the Interior, Washington, DC U.S.A. 

Penteriani, V. AND B. Fatvre. 2001. Effects of harvesting 
timber stands on goshawk nesting in two European 
areas. Biol. Conserv. 101:211-216. 

, M. Mathiaut, and G. Boisson. 2002. Immediate 

species responses to catastrophic natural disturbanc- 
es: windthrow effects on density, productivity, nesting 
stand choice and fidelity in Northern Goshawks {Ac- 
cipiter gentilis) . Auk 119:1132—1137. 

Ramsey, F.L., M. McKracken, J.A. Crawford, M.S. Drut, 


and W.J. Ripple. 1994. Habitat association studies of 
the Northern Spotted Owl, Sage Grouse, and Flam- 
mulated Owl. Pages 189-209 in N. Lange, L. Ryan, L. 
Billard, D. Brillinger, L. Conquest, and J. Greenhouse 
[Eds.] , Case studies in biometry. John Wiley and Sons, 
Inc., New York, NY U.S.A. 

Reutebuch, S.E. and B.F. Gall. 1990. Using historical 
aerial photos to identify long-term impacts of forestry 
operations in sensitive watersheds. Pages 112-120 in 
Proceedings of ACSM-ASPRS fall convention 5-10 No- 
vember 1990, Anaheim, CA. American Society for 
Photogrammetry and Remote Sensing; Am. Congress 
on Surveying and Mapping. Bethesda, MD U.S.A. 

Reynolds, R.T. 1975. Distribution, density, and produc- 
tivity of three species of accipiter hawks in Oregon. 
M.S. thesis, Oregon State Univ., Corvallis, OR U.S.A. 

. 1978. Food and habitat partitioning in two 

groups of coexisting Accipiter. Ph.D. dissertation, 
Oregon State Univ., Corvallis, OR U.S.A. 

. 1983. Management of western coniferous forest 

habitat for nesting accipiter hawks. USDA Forest Ser- 
vice, Rocky Mtn. For. and Range Exp. Sta., Gen. Tech. 
Rep. RM-102, Fort Collins, CO U.S.A. 

, R.T. Graham, M.H. Reiser, R.L. Bassett, P.L. 

Kennedy, D.A. Boyce, Jr., G. Goodwin, R. Smith, and 
E.L. Fisher. 1992. Management recommendations for 
the Northern Goshawk in the Southwestern United 
States. USDA Forest Service, Rocky Mountain Forest 
and Range Experiment Station. Gen. Tech. Rep. RM- 
217, Ft. Collins, CO U.S.A. 

and S.M. Joy. 1998. Distribution, territory occu- 
pancy, dispersal, and demography of Northern Gos- 
hawks on the Kaibab Plateau, Arizona. Unpublished 
final report for Arizona Game and Fish, Heritage Pro- 
ject Number 194045. USDA Forest Service, Rocky 
Mountain Research Station, Fort Collins, CO U.S.A. 

, , AND D.G. Leslie. 1994. Nest productivity, 

fidelity, and spacing of Northern Goshawks in Arizo- 
na. Stud. Avian Biol. 16:106-113. 

, E.C. Meslow, and H.M. Wight. 1982. Nesting 

habitat of coexisting Accipiter m Oregon./. Wildl. Man- 
ag. 46:124-138. 

Rosenfield, R.N. and J. Bielefeldt. 1996. L i f etime nest- 
ing area fidelity in male Cooper’s Hawks in Wisconsin 
Condor 98:165-167. 

Rosenzweig, M.L. 1985. Some theoretical aspects of hab- 
itat selection. Pages 517-540 in M.L. Cody [Ed.], Hab- 
itat selection in birds. Academic Press, Inc., Orlando, 
FL U.S.A. 

SAS Institute, Inc. 1992. SAS procedures guide for per- 
sonal computers, 7* ed., Cary, NC, U.S.A. 

SiDERS, M.S. AND P.L. Kennedy. 1996. Forest structural 
characteristics of Accipiter nesting habitat: is there an 
allometric relationship? Condor 98:123-132. 

Smallwood, K.S. 1998. On the evidence needed for list- 
ing Northern Goshawks {Accipiter gentilis) under the 


September 2005 


Conservation 


323 


Endangered Species Act: a reply to Kennedy. J. Raptor 
Res. 32:323-329. 

Squires, J.R. and L.F. Ruggiero. 1996. Nest-site prefer- 
ence of Northern Goshawks in southcentral Wyo- 
ming./. Wildl. Manag. 60:170-177. 

U.S. Department of Agriculture, Forest Service. 1989. 
Fremont National Forest, Land and Resource man- 
agement plan. Lakeview, OR U.S.A. 

. 1993. Eastside forest ecosystem health assess- 
ment, Vols. I— rV. USDA Forest Service, Pacific North- 
west Region. Portland, OR U.S.A. 

. 1994. Revised decision notice for the continua- 
tion of interim management direction establishing ri- 
parian, ecosystem and wildlife standards for timber 
sales. Regional Forester’s Plan Amendment #1, Re- 
gion 6, east of the Cascade crest in Oregon and Wash- 
ington. USDA Forest Service, Pacific Northwest Re- 
gion. Portland, OR U.S.A. 20 May 1994. 

Urban, D.L. and H.H. Shugart, Jr. 1986. Avian demog- 
raphy in mosaic landscapes: modeling paradigm and 


preliminary results. Pages 273—279 inj. Verner, M.L. 
Morrison, and C.J. Ralph [Eds.], Wildlife 2000: mod- 
eling habitat relationships of terrestrial vertebrates. 
Univ. Wisconsin Press, Madison, WI U.S.A. 

Vales, D.J. and F.L. Bunnell. 1985. Comparison of meth- 
ods for estimating forest overstory cover. Research, 
Ministries of Environment and Forests, Victoria, Brit- 
ish Columbia, Canada. 

WOODBRIDGE, B. AND P. Detrich. 1994. Territory occu- 
pancy and habitat patch size of Northern Goshawks 
in the southern Cascades of California. Stud. Avian 
Biol. 16:83-87. 

, , AND P.H. Bloom. 1988, Territory occu- 
pancy, nest site and habitat patch relationships, and 
movement by nesting Northern Goshawks: implica- 
tions for management. Western Section of The Wild- 
life Society, Hilo, HI U.S.A. 

Received 4 March 2004; accepted 20 May 2005 

Guest Editor: Patricia L. Kennedy 

Associate Editor: Clint Boal 


J Raptor Res. 39(3):324-334 
© 2005 The Raptor Research Foundation, Inc. 

MONITORING RESULTS OF NORTHERN GOSHAWK NESTING 
AREAS IN THE GREATER YELLOWSTONE ECOSYSTEM: 

IS DECLINE IN OCCUPANCY RELATED TO HABITAT CHANGE? 

Susan M. Patla^ 

Northern Rockies Conservation Cooperative, RO. Box 250, Jackson, WY 83001 U.S.A. 

Abstract. — I monitored a subset of Northern Goshawk {Accipiter gentilis) nesting areas on the Targhee 
portion of the Caribou-Targhee National Forest in eastern Idaho and western Wyoming from 1998- 
2002 (recent period) to provide occupancy and productivity data for U.S. Forest Service monitoring 
requirements. A total of 16 randomly-selected nesting areas, half in undisturbed and half in timber-sale 
project areas, were surveyed each year. Occupancy in 1998-2002 averaged 34%, which was significantly 
lower than the 61% measured at these nesting areas from 1992-95 (baseline period) using similar survey 
methods and effort. Productivity of successful nests was similar between the two periods. I used the 
dawn vocalization survey method in 2001-02, in addition to standard broadcast survey methods, to 
determine if low occupancy reflected a poor detection rate of pairs that occupied sites, but failed to 
reproduce. Detection rate of goshawks during the courtship period in these 2 yr averaged less than 
50%, indicating that number of pairs reoccupying known nesting areas surveyed was low. I found no 
relation between weather factors and lower occupancy. Occupancy at nesting areas located in past tim- 
ber-harvest areas in the recent period was significantly lower compared to those in less disturbed habitat 
(22% occupancy versus 45%, respectively) suggesting that occupancy may be influenced by the long- 
term effects of timber-management practices. Whether the observed decline during the recent period 
reflects spatial shifts of nesting pairs, short-term demographic responses to variation in weather or prey, 
or longer-term responses to changes in forest structure and age resulting from timber-management 
activities, cannot be determined using the current monitoring program. Long-term monitoring of study 
areas in the western United States, based on statistically valid study designs and adequate sample size, 
is needed to understand if the apparent decline in goshawk occupancy reported here and in other 
recent studies has serious implications for conservation of this species. 

Key Words; Northern Goshawk; Accipiter gentilis; nest-site occupancy; raptor monitoring, survey techniques; 
forest management. 


RESULTADOS DEL MONITORED DE AREAS DE NIDIFICACION DE ACCIPITER GENTILIS EN EL 
AMPLIO ECOSISTEMA DE YELLOWSTONE: ^ESTA RELACIONADA LA DISMINUCION EN LA OCU- 
PACION CON EL CAMBIO DEL HABITAT? 

Resumen. — Evalue un conjunto de areas de nidificacion de Accipiter gentilis en la porcion Targhee del 
Bosque Nacional de Caribou-Targhee en el este de Idaho y oeste de Wyoming desde 1998 hasta 2002 
(periodo actual) para proveer datos de ocupacion y productividad para los requerimientos de evaluacion 
del Servicio Forestal. Un total de 16 areas de nidificacion seleccionadas al azar fueron evaluadas cada 
aho (la mitad en areas no perturbadas y la mitad en areas de proyectos de venta de madera). La 
ocupacion promedio durante el periodo actual fue de un 34%, lo cual fue significativamente menor 
que el 61% medido en areas de nidificacion desde 1992 hasta 1995 (periodo de linea de base) usando 
metodos y esfuerzos de muestreo similares. La productividad de los nidos exitosos fue similar entre los 
dos periodos. Realice muestreos de vocalizaciones durante el amanecer en 2001 y 2002, ademas de otros 
metodos estandar de reproduccion de grabaciones, para determinar si la baja ocupacion reflejaba una 
tasa de deteccion baja de las parejas que ocupaban los sitios pero que no se reproducfan. La tasa de 
deteccion de A. gentilis durante el periodo de cortejo en estos dos anos fue en promedio menos del 
50%, indican do que el numero de parejas que ocuparon nuevamente las areas conocidas de nidificacion 
fue bajo. No encontre una relacion entre los factores climaticos y una baja ocupacion de individuos. La 
ocupacion durante el periodo actual en las areas de nidificacion en las que se cosecho madera en el 
pasado fue significativamente menor comparada con la de arabientes menos perturbados (22% de 


^ Email address: susan. patla@wgf. state .wy. us 


324 


September 2005 


Conservation 


325 


presencia comparado con 45%, respectivamente), sugiriendo que la ocupacion podria estar asociada 
con los efectos de largo plazo de las practicas de manejo forestal. Usando el plan actual de evaluacion, 
no es posible determinar si la disminucion observada durante el periodo actual refleja desplazamientos 
espaciales de parejas nidificantes, respuestas demograficas de corto plazo a la variacion en el clima o 
las presas, o respuestas de largo plazo a los cambios en la estructura del bosque y en las clases de edad 
resultantes de las actividades de manejo forestal. Es necesaria una evaluacion a largo plazo de las areas 
de estudio en el oeste de los Estados Unidos, basada en estudios con disenos estadisticamente validos 
y tamanos de muestreo adecuados, para entender si la disminucion aparente en la ocupacion de A. 
gentilis presentada aqui y en otros estudios recientes tiene implicancias serias para la conservacion de 
esta especie. 

[Traduccion del Equipo Editorial] 


Concern over potential effects of forest manage- 
ment on Northern Goshawk {Accipiter gentilis) pop- 
ulations nesting in western North America has 
stimulated research on this species since the early 
1970s (Squires and Reynolds 1997). The U.S. For- 
est Service (USFS) controls a large proportion of 
forested lands in the western United States, and 
how forest habitat is managed on these lands has 
been a primary focus of past goshawk research 
(Reynolds 1983, Crocker-Bedford 1990, Reynolds 
et al. 1992). The goshawk is classified as a Sensitive 
Species and a Management Indicator Species for 
forested habitats on the Caribou-Targhee National 
Forest (CTNF) where this study was conducted 
(USDA 1997a). The USFS is required to monitor 
goshawk population trend and its relationship to 
habitat change for designated Management Indi- 
cator Species by federal regulations resulting from 
implementation of the National Forest Manage- 
ment Act of 1982. 

Little information existed on goshawk nesting 
ecology or habitat on the CTNF prior to the 1990s. 
From 1989-95, I conducted surveys and collected 
data on demographic and habitat parameters at 
four historic and 27 occupied nesting areas located 
in a variety of habitats and management areas 
across the forest (Patla 1997). In 1997, the CTNF 
adopted a revised Land Management Plan (LMP) 
that required monitoring a minimum of 15 ran- 
domly-selected nesting areas for adult occupancy 
each year as an indicator of population trend 
(USDA 1997b). I conducted these surveys annually 
for occupancy and productivity from 1998-2002. 
To provide some insight on potential associations 
between timber harvest and resultant habitat 
change on goshawk demographics, I selected 16 
nesting areas each year: half located within past 
timber-sale project areas and half from relatively 
undisturbed areas. 

The objectives of this study were: (1) to compare 


demographic data collected from 1998-2002 (re- 
cent period) to comparable data collected during 
a baseline-study period from 1989-95, as an indi- 
cation of population trend of known nesting areas, 
(2) to compare demographic data collected at 
nesting areas in relatively undisturbed habitat to 
those in timber-harvest management areas to ex- 
amine if goshawk occupancy patterns changed re- 
lated to timber-harvest activities, (3) to provide in- 
formation on survey methods and results including 
a description of a dawn-vocalization survey, and (4) 
to discuss implications of this study and the need 
to improve monitoring efforts in study areas in the 
Intermountain West. 

Study Area 

The Targhee portion of the CTNF contains ca. 728 000 
ha in eastern Idaho and western Wyoming and comprises 
the western portion of the Greater Yellowstone Ecosystem 
(GYE) as described by Clark and Zaunbrecher (1987; Fig. 
1). Most of the CTNF falls within the Middle Rocky 
Mountain physiographic province except for a small por- 
tion, which is included in the Northern Rocky Mountain 
Province (Steele et al. 1983). Elevations range from 
1585-3470 m. The climate is characterized by long, cold 
winters with heavy snowfall and mild, dry summers. Mean 
temperatures are —8° and 18°C for January and July, re- 
spectively, and total annual precipitation ranges between 
61 and 102 cm (Patla 1997). 

Douglas-fir {Pseudotsuga menziesii van glauca) and lod- 
gepole pine {Pinus contorta) are the most common coni- 
fer species within the montane zone, between 1800-2500 
m (Habeck 1994), and are the primary commercial tree 
species harvested on the forest. The dominant cover type 
at 31 goshawk nesting areas within a radius of 2428 ha 
centered at known nest trees was Douglas-fir {N = 14), 
mixed conifer {N = 9), and lodgepole pine {N = 8; Patla 
1997). I found the majority of goshawk nests {N = 49) 
in mature Douglas-fir (78%) and lodgepole pine (8%) 
trees. Mean age of Douglas-fir and lodgepole pine trees 
used for nesting was 143 yr and 96 yr, respectively. 

The CTNF initiated a commercial timber sale program 
in the early 1960s, and an estimated 1935 million board 
feet (MBF) of mature timber was harvested from 1963- 
2001 (Fig. 2; M. Jenkins, CTNF Silviculturist, unpubl. 
data). The mean annual harvest was 62.0 MBF (1963—92) 


326 


Patla 


VoL. 39, No. 3 



Figure 1. Location of the study area of the Caribou-Targhee National Forest in the Greater \fellowstone Ecosystem 
in relation to adjacent national forests and parks. 


but dropped to 8.2 MBF in recent years (1993-2001). 
Harvest methods included clear-cutting of lodgepole 
pine and seed tree or shelterwood cuts of Douglas-fir (Pa- 
tla 1997). No large-scale timber harvesting projects oc- 
curred in the vicinity of known goshawk nesting areas 
during the current study period. 

The revised 1997 CTNF Land Management Plan man- 
ages goshawk nesting habitat by specifying the level, type, 
and timing of management activities that can be con- 
ducted at different spatial areas surrounding historical 
and current nesting areas (USDA 1997b). Prior to the 
late 1990s, a few occupied goshawk nests found in timber 
sale units were protected by creation of small buffers (a 
few trees up to 4 ha; Patla 1997) . The majority of har- 
vesting occurred on the CTNF prior to the implementa- 
tion of goshawk monitoring protocols. 

Methods 

Sampling Unit and Scheme. The sampling unit moni- 
tored in during the recent period, 1998-2002, was the 
nesting area which included all known nests used by a pair 
of goshawks and the surrounding area of 1.6 km radius 
measured from a centroid based on known nest locations 
(Woodbridge and Detrich 1994, Siders and Kennedy 
1996). The size of the defined nesting area (2428 ha) 
was based on known nearest-neighbor distance data and 


territory spacing measured in this study area and others 
in the western United States and should be sufficient to 
distinguish between nesting pairs (Reynolds et. al. 1994, 
Woodbridge and Detrich 1994, Patla 1997). 

I monitored 16 nesting areas each year, randomly se- 
lected from a master list of 34, that had been occupied 
by a pair of goshawks at least once since 1989. I excluded 
from the selection process a few nesting areas in difficult 
to access locations, and also some historical nesting areas 
occupied prior to 1989 in which harvesting had subse- 
quently eliminated known nest stands, and where I had 
found no evidence of goshawk use during the baseline 
study (1989-95). Prior to selecting monitoring sites, I 
classified the 34 nesting areas into one of two categories: 
(1) undisturbed sites located outside of the boundaries 
of timber-management project areas (N = 15) with little 
or no harvesting within the defined nesting area, or (2) 
timber-harvest sites {N = 19). Eight nesting areas from 
each category were monitored each year 1998-2002. 

I included in the timber-harvest category all goshawk 
nesting areas with nest sites that fell within the bound- 
aries of past timber sale projects. Thus, timber-harvest 
sites included a range of disturbance conditions. I did 
not quantify differences between undisturbed and tim- 
ber-harvest sites as a detailed vegetation analysis of nest- 
ing areas was beyond the scope of the current study. 


September 2005 


Conservation 


327 




Year 


Figure 2. Timber harvest activity on the Targhee section of the Caribou-Targhee National Forest showing volume 
cut in million board feet per year, 1964-2002. Northern Goshawk monitoring was initiated in 1989. 


Based on previous analysis of ten nesting areas found in 
timber-sale project areas and subsequently harvested be- 
tween 1985-92, harvesting resulted in a reduction in ma- 
ture forest cover within the defined nesting area (Patla 
1997). Prior to harvest, mature forest habitat averaged 
80% (range = 63-95%) within the nesting area com- 
pared to 61% (range = 51-80%) post-harvest. Reduction 
of forest habitat was greatest in the center of the nesting 
area (see Patla 1997). 

To compare monitoring results from the recent period 
to the baseline period, I first removed data for baseline 
sites not surveyed at the same spatial scale using the 
broadcast survey method (see Patla 1997 for description 
of baseline survey methods) . I then made a random se- 
lection of 16 sites from those years, 1992-95, for which I 
had a sample pool greater than 16 that met sampling 
criteria. I did not include results from 1990-91 due to 
inadequate number of nesting areas and from 1996-97 
because I did not monitor nesting goshawks during these 
years. 

Monitoring Terms. I considered a nesting area occu- 
pied if: (1) a pair was observed vocalizing, copulating, or 
nest building in the vicinity of known nests, or a single 
adult was heard vocalizing in the vicinity of a known nest 
tree during the pre-nesting period on more than one day 
(mid-March-early May), (2) a single adult or pair de- 
fended a nest site during the incubation/nestling period, 
or evidence of nest building or egg laying was confirmed 


(late April-early June), or (3) young of the year were 
found during the nestling (June-mid-July) or fledgling 
periods (July-mid-August; Postupalsky 1973, Steenhof 
1987, Woodbridge and Detrich 1994). I classified a nest- 
ing area as not occupied if a single adult was heard mak- 
ing a few calls on only one day during the courtship pe- 
riod, or if an adult was seen on a single occasion later in 
the season, and no additional signs of goshawk presence 
were detected within the 1.6-km radius survey area. I clas- 
sified a pair as having laid eggs and attempted to nest 
when an adult was found incubating or young were ob- 
served in the nest (Postupalsky 1973). Nests were classi- 
fied as successful if at least one fledgling or fully-feath- 
ered nestling (ca. 5-wk post hatching) was observed. Each 
year that a nesting area was monitored was considered a 
territory year for calculating occupancy rates (Wood- 
bridge and Detrich 1994). 

Survey Methods. I based timing of surveys in the re- 
cent period on nesting chronology calculated from 37 
successful nest attempts, 1989-94 (Patla 1997). I used 
similar survey methods during both time periods. The 
mean onset of incubation was 5 May (range = 20 April- 
20 May) , mean hatching date was 6 June (range = 22 
May-June 21), and the mean fledge date was 15 July 
(range = 1 July-3 August) . 

All known nest trees and stands were checked visually 
in April or May for goshawk activity (e.g., fecal deposits, 
molted feathers, conifer sprigs on nests). If pairs were 


328 


Patla 


VoL. 39, No. 3 


not detected, standardized broadcast calling surveys 
(Kennedy and Stahlecker 1993, Joy et al. 1994) were con- 
ducted in forest habitat within a 0.8-km radius of the last 
used nest during the nestling period (early June— mid- 
July) . Survey effort was expanded in the fledgling period 
(mid-July-end of August), if no detections were obtained 
on earlier surveys, to cover a 1.6-km radius area based on 
a centroid of known nests. Transect lines were 260 m 
apart, and calling stations ranged from 150-300 m, de- 
pending upon terrain and density of forest cover. Surveys 
were not conducted on days when wind or rain interfered 
with the ability to transmit calls or hear detections. Oc- 
cupied nests were monitored every 2 wk to determine 
number of young and approximate fledging date. 

In 2001 and 2002, in addition to broadcast surveys dur- 
ing the nestling and fledgling periods, 1 surveyed all se- 
lected nesting areas using the dawn vocalization survey 
method to increase the likelihood of detecting pairs that 
abandoned the nesting effort early in the season, prior 
to egg laying or incubation (Penteriani 1999, Dewey et 
al. 2003). Observation points were selected within 100- 
200 m from the last nest tree occupied, or centered be- 
tween clusters of alternate nest trees located within a few 
hundred meters of each other. At nesting areas with nests 
located >200 m apart, either two observers stationed 
within 100-200 m of known nests were used, or a single 
observer completed surveys on different days. Observa- 
tion periods lasted for a minimum of 2 hr starting 30-45 
min prior to sunrise and ending 1.5-2 hr after sunrise. 
If no goshawk activity was detected during the initial sur- 
vey, follow-up surveys were conducted 1-3 wk later, if pos- 
sible. 

Statistical Analysis. To compare demographic results 
between the baseline and recent monitoring periods, and 
undisturbed and timber harvest nesting areas in the re- 
cent monitoring period, I treated the 16 nesting areas 
selected each year as independent samples. 

For most statistical analyses, I applied a multi-response 
permutation process (MRPP) that is analogous to one- 
way analysis of variance (or t-test), using Blossom 
(Cade and Richards 2001). MRPP statistical procedures 
have no distribution assumptions and work well for eco- 
logical data with small sample sizes that lack normal dis- 
tribution even after data transformations (Cade and 
Richards 2001). I used the chi-square contingency test to 
compare number of occupied nesting areas in undis- 
turbed and timber sale areas in the recent period. Sig- 
nificance level for all tests was P = 0.05. 

Analysis of Weather Parameters. Weather factors have 
been shown to influence occupancy of goshawk nests 
(Kostrzewa and Kostrzewa 1991, Patla 1997). To analyze 
potential effects of drought, I compared total annual pre- 
cipitation between the baseline and recent periods in- 
cluding the year prior to each defined time segment 
based on precipitation measured at Driggs, ID (Climate 
Station No. 2676, Teton County, elevation 1866 m) near 
the center of the study area (Idaho State Climate Services 
2002). I also compared snowwater equivalents (SWE) in 
March between the baseline and recent periods (Pine 
Creek Pass, Climate Snow Station No. 6720, Teton Coun- 
ty, elevation 2049 m) (Idaho State Climate Services 
2002). SWE is computed from snow density to determine 
percent water content in the snow pack. 


Results 

Survey Effort. Eighty territory-year checks were 
completed (16 nesting areas monitored/yr) during 
the current monitoring period. Thirty of the 34 
nesting areas (88%) on the master list were mon- 
itored at least once. Sampling frequency ranged 
from 0-5 yr (x = 2.4 yr) for individual nesting ar- 
eas. 

Observers visited nesting areas on average 5.7 ± 
0.87 (SD) times per breeding season. At nesting 
areas where no occupied nests or young were de- 
tected, observer effort averaged 64 ± 18 (SD) call- 
ing stations, and 15.9 ± 7.9 hr/ territory in suitable 
habitat. Similar effort was expended during 1992- 
95 with calling stations played within a 1.6-km ra- 
dius at similar intervals using the same protocol. 

Occupancy and Productivity. The mean occu- 
pancy rate in the recent period was 34% and was 
significantly lower compared to the baseline period 
(61%; MRPP: P = 0.031; Table 1). Occupancy rate 
was highest in 1992 and then declined in subse- 
quent years (Table 1, Fig. 3). 

In the current monitoring period, 20% of nest- 
ing areas had successful nests and produced a total 
of 35 young (Table 1). Nest success and total num- 
ber of young produced was significantly higher 
(MRPP: P = 0.003 and P = 0.004) during 1992-95 
(Table 1). However, mean number of young per 
nesting pair and per successful nest was nearly 
identical between the two study periods (Table 1). 

Weather Analysis. I found no significant differ- 
ence in two weather factors analyzed that might 
have influenced comparative occupancy rates. 
Mean annual precipitation was similar between the 
1992-95 period (30.9 ± 3.5 cm) and recent period 
(1996-2002; 36.6 ± 7.1 cm; MRPP, P = 0.109). 
March SWE was also similar: 32% in 1992-95 com- 
pared to 33% in 1996-2002 period (MRPP, P — 
0.928). 

Undisturbed Versus Timber-harvest Nesting Ar- 
eas. In the recent period (1998—2002), a signifi- 
cantly greater number of undisturbed nesting 
areas (18/40 = 45 ± 14%) compared to timber- 
harvest nesting areas (9/40 = 22.5 ± 10%) were 
occupied (x^, P — 0.033; Table 2). Pairs in undis- 
turbed nesting areas produced a greater number 
of young per yr (Table 2, MRPP: P = 0.027). The 
mean number of young produced per nesting pair 
and per successful territory was similar (Table 2) . 

Mean occupancy rates during the 1996-2002 
monitoring period at both undisturbed (45.0%) 


September 2005 


Conservation 


329 


4-T 

d 

V 


u 


0 

15 


> 

Sh 


b 

cd 

cn 

G 

to 

0 

C4 

'O 

oi 

Z 

IS 

■l-l 


CO 

V 

D 

V 

b 


Cm 

bJO 

u 

O 


b 

H 

G 

15 

D! 

b 


•C i= 

6 3 

D B 

" § 

0 

CJ 

01 ^ 

^ pC 

C/5 

^ trt 

V f3 

b 

Cm 

WD 

bc u 
H U 
.in O 

b 

b t 3 
^ !=l 

rt 

.S 

CA -4_) 

O 

O^ 
!=i ^• 

^ in 

V CD 

J. 

■“ O'! 
O C5 

Z ^ 

■■rt "b 

V 

u 

'o 
u 

n 

-4-J 

rt 
O t3 


(j 

u 

Ij 

fJl 


Tj 

G 

cj 

u 


a 

o 


V 

fl 


rt 


(M 

s 

■1-1 O 
Si cv( 

J, 

bc cn 
b b! 

■c 

•- 

b P 


o 

a 

IfH 

o 


Sh 

V 

fX 


b 

< Si 
H Si 


si 

a 

a 

b 

cn 


u 


b 

u 

V 

jh 

■w 

bo 

b 

• pn 

u 

b 

a 


J 

p 

b 

c/5 

C/5 

W 

u 

u 

p 

C/D 


pc! 

W 

Pn 

o 

z 

p 


Pi 

b 

PLh 

O 

Z 

P 


Q 

w 

S 

P 

u 

u 


;20^ 


w e 
o 
o 
p 
cn 




o 

z 

)— < 

b 

z 


u 


Pm 

P 

o 

u 

O 


o 

H 






b 

o 

pc! 

W 

PP 

s 

p 

z 


o 

z 

p 


C/5 

O 

z 

Q 

5 “ 

Ph [i( 






o 

e ^ 
Z 



00 r-H t-H r-H 

O 1> l> 1— I T-1 


CM 


ffvf (M Csf o 


I> O ii ®o o 
CD in 1 > rf^ 00 CD 

T— I 1-H CcT 


O ^ Sf CO O 

CD in i> 00 CD 

T“H 11 Ijvj 1“H 1“H 1^) 


m ^ ^ 00 o 00 

i> cn in Ml 

O O <3 O O O 


Tf o CD 1-H 1 > 

05 in in CD CM 

o o o o o o 


'tH O CD 1 — I 

05 m -bi m cc cif 

d d o d o d 


en in 

m 05 on .!> d 
Gcf r1 ^ — ' 


q q 

CM i> CO 00 ni 


CC TfH 

m 00 i> 05 d 


o o m m on (M 
m q q q q 1 — ; q 

CM CM CM iM d d d 


o o in in !>• 1— I m 
q CM CM q q q 
CM O CM d ^ O 


O O O COS csn CD CM 

in q q q q q q 

l^b C ^ C— C ^b 


in CD in in 05 o 00 

CM O CM CM iH CM o 

<6 d di CP <5 d d 


m th m m 05 m 

q q q q rH q q 

d d d d d d d 


in ii 00 'cf x> 

CM on on 'cfi on sn o 

CC CP <6 d> <6 <6 CD 


q q 

o CM 05 05 m d on 


q q 
if on on ^ 


o 

if m if if on If d 


CD 


CD 


q q 

m 00 i> C35 d d 


q 

m m .!> CD in 


CM on if m 

C35 05 05 05 
05 05 05 05 


§ 

o 


Q 

C/D 


b 

<u 

CJ 

Pi 


00 05 
05 05 
05 05 


O 1H CM 

o o o 
o o o 

CM CM CM 


§ 

CU 


330 


Patla 


VoL. 39, No. 3 



Year 

Figure 3. Regression analysis of occupancy rate 1992-2002 of Northern Goshawk nesting areas on the Caribou- 
Targhee National Forest. Data not available for years 1996 and 199V. 


and timber-harvest nesting areas (22.5%) were low- 
er compared to the 61% mean occupancy rate 
measured 1992-95 (Table 1). The observed de- 
cline in the recent period was over twice as high 
in timber-harvest nesting areas (63%) compared to 
undisturbed nesting areas (26%). 

Use and Spacing of Alternate Nest Trees. Of 18 
nest attempts documented from 1996-2002, only 
three (17%) occurred in previously identified nest 
trees. Eighty percent (N — 12) of new alternate 
nest trees found were located within 800 m of the 
last nest tree used. Mean distance between alter- 
nate nest trees used within a nesting area was 572 
± 352 m, (range — 77-1381 m, N = 15). I found 
no relationship in the distance between nest trees 
used and the span of years since goshawks had 
been documented nesting in a particular area 
(Pearson correlation, r = 0.009, P = 0.975). Nest 
trees tended to be clustered within the defined 
nesting areas. 

Goshawk Detections and the Dawn Survey Meth- 


od. The majority of initial detections confirming 
occupancy of nesting areas (63%, 17/27) occurred 
during the courtship period compared to 26% in 
the nestling (N — 7) and 1 1 % in the fledgling pe- 
riods (N = 3). Most detections in the courtship 
period (71%) were obtained using the dawn vo- 
calization survey method in 2001-02, which result- 
ed in detections of pairs or single adults at 15 sites 
(Table 3). 

Number of dawn surveys completed averaged 
3.0 per nesting area in 2001 (range = 1-5) and 1.9 
(range = 1-3) in 2002. During these two years, 
92% of all nesting areas confirmed occupied re- 
sulted from use of dawn surveys with only one false 
negative at a site where an occupied nest was found 
later in the fledgling period 655 m from the last 
used nest tree. At three sites, detections were doc- 
umented only on one day, with no other goshawk 
activity being observed during the remainder of 
the nesting season. I did not classify these nesting 
areas as occupied (Table 3) given the brevity of the 


September 2005 


Conservation 


331 


Table 2. Comparison of monitoring data from randomly-selected Northern Goshawk nesting areas in undisturbed 
and past timber-harvest areas, Caribou-Targhee National Forest (1998-2002). 


War 

Occupied 

Nesting 

Area 

Nesting 

Pairs 

Successful 

Pairs 

Total 
Number of 
Young 

Young per 
Occupied 
Nest Area 

Young per 
Nesting Pair 

Young per 
Successful 
Nest Area 

Undisturbed 

1998 

{N= 8/yr) 
3 

3 

3 

7 

2.33 

2.33 

2.33 

1999 

2 

0 

0 

0 

0.00 

0.00 

0.00 

2000 

4 

3 

3 

7 

1.75 

2.33 

2.33 

2001 

5 

3 

3 

8 

1.60 

2.67 

2.67 

2002 

4 

2 

2 

4 

1.00 

2.00 

2.00 

Mean 

3.6 

2.6 

2.2 

5.2 

1.34 

1.87 

1.87 

(SD) 

(1.1) 

(0.5) 

(1.3) 

(3.3) 

(0.9) 

(1.1) 

(1.1) 

Timber-harvest 

1998 

(A= 8/yr) 

1 

1 

1 

3 

3.00 

3.00 

3.00 

1999 

3 

3 

1 

2 

0.67 

0.67 

2.00 

2000 

1 

1 

1 

2 

2.00 

2.00 

2.00 

2001 

2 

1 

1 

1 

0.50 

1.00 

1.00 

2002 

2 

1 

1 

1 

0.50 

1.00 

1.00 

Mean 

1.8 

1.4 

1.0 

1.8 

1.33 

1.53 

1.80 

(SD) 

(0.8) 

(0.9) 

(0.0) 

(0.8) 

(1.10) 

(1.00) 

(0.80) 


vocalizations and lack of other evidence confirm- 
ing occupancy. Without use of the dawn survey 
method, 43% of nesting areas in 2001 (N = 3) and 
33% in 2002 {N = 2) would not have been classi- 
fied as occupied. However, even with use of court- 
ship surveys, occupancy of nesting areas during 
these two years fell below the 61% average from 
1992-95 (Table 1). Occupancy in 2001, the highest 
occupancy year during 1996-2002, was the same 
(44%) as 1994, the lowest occupancy year from 
1992-95. Even if goshawk pairs or individuals were 
detected during the courtship period, follow-up 
broadcast calling surveys were required at many 
sites later in the season to locate an occupied nest 
tree and determine number of young. 


Discussion 

Monitoring Effort and Study Design. Monitoring 
Northern Goshawk nesting populations is chal- 
lenging given the secretive nature of the species, 
its use of widely-spaced alternate nests, spatial and 
temporal variability in numbers of nesting pairs, 
and density of the mature forest habitat used for 
nesting (Woodbridge and Detrich 1994, Kennedy 
1997, Squires and Reynolds 1997). Comparison of 
occupancy among studies is also difficult as occu- 
pancy estimates appear to be positively correlated 
with amount of effort expended to determine nest- 
ing area status (Kennedy 1997). Multiple visits to 
nesting areas over the course of a season and 
broadcast-calling surveys at least up to 1.6-km ra- 


Table 3. Results of dawn vocalization surveys during March and April at Northern Goshawk nesting areas {N = 16/ 
yr) on the Caribou-Targhee National Forest (2001-02). 


War 

Total No. 
Surveys 

EuARLIEST 

Detection 

Date 

Latest 

Detection 

Date 

No. 

Detections^ 

(Occupancy) 

No. Detections 
Single Bird*’ 

Error 

Rate‘s 

2001 

48 

31 Mar 

2 May 

7 (0.44) 

1 

0.00 

2002 

30 

26 Mar 

20 April 

5 (0.31) 

2 

0.06 

Total 

78 

26 Mar 

2 May 

12 (0.38) 

3 

0.03 


“ Number of territories classified as occupied where a pair was detected or a single adult was heard or seen on more than 1 day 
^ Number of territories where detections consisted only of a few “kek" calls heard briefly one day. 

Error rate defined as the proportion of territories misclassified as unoccupied and later confirmed as occupied. 


332 


Patla 


VoL. 39, No. 3 


dius around nest sites are required to monitor pre- 
viously identified nesting areas effectively (Reyn- 
olds et al. 1994, Woodbridge and Detrich 1994, 
Finn et al. 2002) . 

The amount of survey effort expended per nest- 
ing area (mean number of visits per site and area 
surveyed) for the current goshawk monitoring pe- 
riod matches or exceeds that reported in other 
long-term goshawk studies (Kennedy 1997, Boyce 
et al. 2005, Reynolds et al. 2005). The total number 
of nesting areas monitored per year was relatively 
low, however, and included only a subset of known 
areas scattered throughout the CTNF. 

The goshawk-monitoring plan for the CTNF is 
based on the assumption that goshawks exhibit ter- 
ritorial behavior and that “a stable population 
should revolve around some average occupancy 
rate” of known nesting areas (USDA 1997b). The 
plan assumes that the occupancy measured at a 
subset of known nesting areas can be used as an 
index of population stability or decrease for the 
species. The plan states: “A sustained downward 
trend of adult occupancy for at least four years may 
indicate a need for action” (USDA 1997b). There 
are no specific requirements that monitoring pro- 
tocols developed for land management plans fol- 
low statistically rigorous study design criteria. The 
approach to monitoring on a forest level tends to 
be pragmatic and based on limited funding avail- 
ability. Whether the study design used on the 
CTNF is adequate as an index for local population 
trend requires further statistical evaluation. For 
this analysis, I assumed that occupancy results ap- 
ply to the target population of known nesting areas 
monitored and may not reflect forest-wide popu- 
lation trends. 

Decline in Occupancy. Results of this study in- 
dicate that occupancy of known goshawk nesting 
areas on the CTNF was significantly higher in the 
early 1990s compared to later in the decade with 
no rebound evident through the 2002 nesting sea- 
son (Fig. 3). Results are consistent with those re- 
ported from other goshawk study areas suggesting 
that there may have been a decline in some gos- 
hawk populations across the Intermountain West 
during the late 1990s (Fairhurst and Bechard 2005, 
Reynolds et al. 2005). 

Results from dawn vocalization surveys on the 
CTNF indicated that the lower occupancy mea- 
sured in the recent period likely did not result 
from failure to detect pairs that occupied sites but 
did not reproduce (Woodbridge and Hargis 2005). 


However, it is possible, given the study design and 
low sample number that spatial shifts by pairs out- 
side of areas surveyed may have confounded re- 
sults. Studies of marked goshawks have shown that 
shifts between nesting areas by individual breeding 
adults occur to some extent and that some ephem- 
eral territories are occupied only occasionally 
(Woodbridge and Detrich 1994, Reynolds and Joy 
1998, Reynolds et al. 2005). If a proportion of pairs 
at study sites on the CTNF shifted each year be- 
tween sites, or used certain sites only occasionally, 
occupancy results could be misleading. 

Weather conditions can influence goshawk oc- 
cupancy, but I did not find a significant difference 
between the recent and baseline periods in rela- 
tion to total annual precipitation and snow water 
equivalents. The latter factor was negatively related 
to occupancy in the baseline study period (Patla 
1997). There may be other local or regional weath- 
er/climatic trends not analyzed in this study that 
were influencing occupancy rates during the study 
period. 

The amount and structure of forest habitat sur- 
rounding nest sites has been related to occupancy 
of historical goshawk nest sites in the western Unit- 
ed States (Crocker-Bedford 1990, Woodbridge and 
Detrich 1994, Desimone 1997, Finn et al. 2002, 
McGrath et al. 2003). During the baseline study 
period, I also found an association between the 
proportion of mature forest habitat and occupancy 
rate of nesting areas on the CTNF. High occupancy 
nest clusters in timber harvest areas, defined as 
those with >50% occupancy rate (N = 16, x oc- 
cupancy = 81%), contained a significantly greater 
proportion of mature forest cover and less young 
forest/seedling cover within a 240 ha area sur- 
rounding known nests compared to low occupancy 
nest clusters (N = 6, occupancy = 37%; Patla 
1997). 

In the recent study period, occupancy of nesting 
areas on the CTNF in timber-harvest areas showed 
a greater proportional decline than those in less 
disturbed habitat, but vegetation differences be- 
tween these categories were not quantified. There 
appears to be an association between reduction in 
mature forest habitat within nesting areas on the 
CTNF as a result of harvesting and decline in oc- 
cupancy. This hypothesis requires further investi- 
gation of vegetation conditions at nest areas in re- 
lation to goshawk occupancy patterns. 

In contrast to occupancy data, I found no differ- 
ence in productivity of nesting goshawk pairs be- 


September 2005 


Conservation 


333 


tween the baseline and recent periods or between 
timber harvest and undisturbed sites in the recent 
period. Pairs that nested successfully produced sim- 
ilar number of young supporting the suggestion by 
McClaren et al. (2002) that number of young 
fledged may not be useful for assessing spatial var- 
iability in goshawk nest habitat quality. 

Whether the decline in occupancy measured at 
known nesting areas on the CTNF reflects a re- 
sponse to decline in quality of primary habitat, a 
shorter-term response to variation in weather and 
prey, or sampling error due to shifting of pairs out- 
side of surveyed sites cannot be determined using 
the current monitoring study plan employed by 
the U.S. Forest Service. However, data from the 
CTNF reflects a pattern documented at other lo- 
cations in the western U.S. that indicated a peak 
in the number of occupied goshawk nest sites in 
1992 and a subsequent decline. It remains un- 
known if goshawk populations exhibit periodic cy- 
clical highs in the western U.S. similar to those 
documented farther north (Doyle and Smith 2001) 
or if trend data indicates the onset of a more se- 
rious, longer-term decline related to habitat or cli- 
matic change. Because many goshawk studies and 
monitoring projects were initiated during or after 
the early 1990s, baseline data prior to 1992 are 
lacking from most areas. How to interpret recent 
trends since 1992 remains challenging. 

Results of the current study emphasize the need 
to develop more comprehensive, well-funded, and 
statistically valid monitoring plans for goshawks 
that can track population trend, reproductive suc- 
cess, and habitat relationships in a timely and 
meaningful way. However, declines at known nest- 
ing areas measured since 1992 suggest that a con- 
servative approach for managing remaining ma- 
ture/old growth forests would be most prudent 
until our knowledge and understanding concern- 
ing the relationship between goshawk demograph- 
ic parameters and loss of mature forest habitat in- 
creases (DeStefano 1998). 

Acknowledgments 

This monitoring project was funded through a Chal- 
lenge-Cost Share agreement with support provided by 
the Caribou-Targhee National Forest and the Northern 
Rockies Conservation Cooperative (NRCC). Idaho Fish 
and Game (IDFG) Nongame Program provided addition- 
al funding. I thank Tim Clark, Denise Casey and the staff 
at NRCC, and Caribou-Targhee NF biologists and staff 
who provided valuable logistical support. This project, 
needless to say, depended upon numerous field assistants 
including Mansori Abe, Mark Berry, Beth Cable, Ned 


Cochran, Melanie Estrella, Sean Finn, Wendy Lammars, 
Frank La Sorte, Don Lingle, Garrett Lowe, Alex Mandel, 
Melissa Merrick, Sophie Osborn, Troy Rintz, Chris Rodes, 
Cari Straight, and Cliff Weiss. I also would like to ac- 
knowledge Dr. Chuck Trost, whose enthusiasm for the 
study of birds brightens the lives of all who know him. I 
thank Sarah Dewey, J. Woodford, and M. Kochert for re- 
viewing a draft of this paper. 

Literature Cited 

Boyce, D.A., Jr., P.L. Kennedy, P. Beier, M.F. Ingraldi, 
S.R. MacVean, M.S. Siders, J.R. Squires, and B. 
WooDBRiDGE. 2005. When are goshawks not there? is 
a single visit enough to infer absence at occupied nest 
areas? J. Raptor Res. 39:296—302. 

Cade, B.S. and J.D. Richards. 2001. Manual for Blossom 
statistical software. Midcontinent Ecological Science 
Center, Fort Collins, CO U.S.A. <http://www.mesc. 
usgs.gov/products/ software/blossom/ shtml/ 

Clark, T. and D. Zaunbrecher. 1987. The Greater Yel- 
lowstone ecosystem: the ecosystem concept in natural 
resource policy and management. Renewable Resources 
Journal. 5:8—16. 

Crocker-Bedford, D.C. 1990. Goshawk reproduction 
and forest management. Wildl. Soc. Bull. 18:262-269. 
Desimone, S.M. 1997. Occupancy rates and habitat rela- 
tionships of Northern Goshawks in historic nesting 
areas in Oregon. M.S. thesis, Oregon State Univ., Cor- 
vallis, OR U.S.A. 

DeStefano, S. 1998. Determining the status of Northern 
Goshawks in the west: is our conceptual model cor- 
rect? J. Raptor Res. 32:342-348. 

Dewey, S.R., P.L. Kennedy, and R.M. Stephens. 2003. Are 
dawn vocalization surveys effective for monitoring 
goshawk nest-area occupancy? J. Wildl. Manag. 67: 
390-397. 

Doyle, F.I. and J.M.N. Smith. 2001. Raptors and scaven- 
gers. Pages 377-404 in C.J. Krebs, S. Boutin, and R. 
Boonstra [Eds.], Ecosystem dynamics of the boreal 
forest. Oxford Univ. Press, New York, NYU.S.A. 
Fairhurst, G.D. and MJ. Bechard. 2005. Relationships 
between winter and spring weather and Northern 
Goshawk {Accipiter gentilis) reproduction in northern 
Nevada. J. Raptor Res. 39:229-236. 

Finn, S.P., D.E. Varland, and J.M. Marzluff. 2002. Does 
Northern Goshawk breeding occupancy vary with 
nest-stand characteristics on the Olympic Peninsula, 
Washington? J. Raptor Res. 36:265-279. 

Habeck, J.R. 1994. Dynamics of forest communities used 
by Great Gray Owls. Pages 176—206 in G.D. Hayward 
andj. Verner [Eds.], Flammulated, Boreal, and Great 
Gray Owls in the United States; a technical conser- 
vation assessment. USDA Forest Service General 
Technical Report. RM-253, Fort Collins, CO U.S.A. 
Idaho State Climate Services. 2002. Annual Reports. 
Biological and Agricultural Engineering Department 
University of Idaho, Moscow, ID U.S.A. 

Joy, S.M., R.T. Reynolds, and D.G. Leslie, 1994. North- 


334 


Patla 


VoL. 39, No. 3 


ern Goshawk broadcast surveys: hawk response vari- 
ables and survey costs. Stud. Avian Biol. 16:24-30. 

Kennedy, P.L. 1997. The Northern Goshawk {Accipiter gen- 
tilis atricapillus) : is there evidence of a population de- 
cline? J. Raptor Res. 31:95-106. 

AND D.W. Stahlecker. 1993. Responsiveness of 

nesting Northern Goshawks to taped broadcasts of 3 
conspecific calls./. Wildl. Manag. 57:249-257. 

Kostrzewa, R. and a. Kostrzewa. 1991. The relationship 
of spring and summer weather with density and 
breeding performance of the buzzard Buteo buteo, gos- 
hawk Accipiter gentilis, and kestrel Ta/co tinnunculus. Ibis 
132:550-559. 

McClaren, E.L., RL. Kennedy, and S.R. Dewey. 2002. Do 
some Northern Goshawk nest areas consistently 
fledge more young than others? Condor 104:343—352. 

McGrath, M.T., S. DeStefano, R.A. Riggs, L.L. Irwin, 
and G.J. Roloff. 2003. Spatially explicit influences on 
Northern Goshawk nesting habitat in the interior Pa- 
cific Northwest. Wildl. Monogr. 154:1-63. 

Newton, I. 1979. Population ecology of raptors. Buteo 
Books. Vermillion, SD U.S.A. 

. 1986. The sparrowhawk. Poyser, Calton, UK. 

Patla, S.M. 1997. Nesting ecology and habitat of the 
Northern Goshawk in undisturbed and timber harvest 
areas on the Targhee National Forest, Greater Yellow- 
stone ecosystem. M.S. thesis, Idaho State Univ., Po- 
catello, ID U.S.A. 

Penteriani, V. 1999. Dawn and morning goshawk court- 
ship vocalizations as a method for detecting nest sites. 
/. Wildl. Manag. 63:511—516. 

. 2001. The annual and diel cycles of goshawk vo- 
calizations at nest sites./. Raptor Res. 35:24—30. 

Postupalsky, S. 1973, Raptor reproductive success: some 
problems with methods, criteria, and terminology. 
Pages 21-31 in F.N. Hamerstrom, Jr., B.F. Harrell, and 
R.R. Olendorff [Fds.], Management of raptors. Rap- 
tor Research Foundation, Vermillion, SD U.S.A. 

Reynolds, R.T. 1983. Management of western coniferous 
forest habitat for nesting accipiter hawks. USD A. For- 
est Service. General Technical Report. RM-102. Rocky 
Mountain Forest and Range Experimental Station, 
Fort Collins, CO U.S.A. 

, R.T. Graham, H.H. Reiser, R.L. Bassett, P.L. 

Kennedy, D.A. Boyce, Jr., G. Goodwin, R. Smith, and 
E.L. Fisher. 1992. Management recommendations for 
the Northern Goshawk in the southwestern United 
States. USDA Forest Service General Technical Re- 
port. RM-217, Fort Collins, CO U.S.A. 


and S.M. Joy. 1998. Distribution, territory occu- 
pancy, dispersal, and demography of Northern Gos- 
hawks on the Kaibab Plateau, Arizona. Final report 
for Heritage Project No. 194045. Arizona Game and 
Fish Department, Phoenix, AZ U.S.A. 

, S.M. Joy, and D.G. Leslie. 1994. Nest productiv- 
ity, fidelity, and spacing of Northern Goshawks in 

northern Arizona. Stud. Avian Biol. 16:106-111. 

, J.D. Wiens, S.M. Joy, C. Erickson, and S.R. 

Salafsky. 2005. Sampling considerations for demo- 
graphic and habitat studies of Northern Goshawks./. 
Raptor Res. 39:274—285. 

SiDERS, M.S. and P.L. Kennedy. 1996. Forest structural 
characteristics of Accipiter nesting habitat: is there an 
allometeric relationship? Cojzdor 98:123-132. 

Squires, J.R. and R.T. Reynolds. 1997. Northern Gos- 
hawk {Accipiter gentilis). Pages 1-31 in A. Poole and F. 
Gill [Eds.], The birds of North America, No. 298. The 
Birds of North America, Inc., Philadelphia, PA U.S.A. 

Steele, R., S.V. Cooper, D.M. Ondov, D.W. Roberts, and 
R.D. Pfister. 1983. Forest habitat types of eastern Ida- 
ho-western Wyoming. USDA Forest Service, Ogden, 
UT U.S.A. 

Steenhof, K. 1987. Assessing raptor reproductive success 
and productivity. Pages 157-170 in B.A. Giron Pen- 
dleton, B.A. Millsap, K.W. Cline, and D.M. Bird 
[Eds.], Raptor management techniques manual: Sci- 
entific and technical series. National Wildlife Feder- 
ation, Washington, DC U.S.A. 

USDA Forest Service. 1997a. Final environmental im- 
pact statement for the Targhee National Forest, Forest 
plan revision. Intermountain Region. Targhee Na- 
tional Forest, St. Anthony, ID 83445 U.S.A. 

USDA Forest Service. 1997b. Revised Forest Plan for the 
Targhee National Forest. Intermountain Region. Tar- 
ghee National Forest, St. Anthony, ID 83445 U.S.A. 

Woodbridge, B. and PJ- Detrich. 1994. Territory occu- 
pancy and habitat patch size of Northern Goshawks 
in the southern Cascades of California. Stud. Avian 
Biol. 16:83-87. 

and C.D. Hargis. 2005. Northern Goshawk in- 
ventory and monitoring technical guide. USDA For- 
est Service, Ecosystem Management Coordination 
Staff, Washington, DC. (http://www.avianscience. 
org/ documents/ Goshawk-landMTech Guide_2005. 
OOO.pdf). 

Received 23 March 2004; accepted 13 June 2005 

Associate Editor: Clint W. Boal 

Guest Editor; John J. Keane 


J Raptor Res. 39(3) :335— 341 
© 2005 The Raptor Research Foundation, Inc. 

EFFECTS OF TIMBER HARVESTING NEAR NEST SITES ON THE 
REPRODUCTIVE SUCCESS OF NORTHERN GOSHAWKS 

{ACCIPITER GENTILIS) 

Todd Mahon' 

University of Alberta/Wildfor Consultants Ltd., P.O. Box 562, Telkwa, British Columbia VOJ 2X0 Canada 

Frank I. Doyle 

Wildlife Dynamics, P.O. Box 129, Telkwa, British Columbia VOf 2X0 Canada 

Abstract. — ^We assessed the effects of timber harvesting near nest sites on the reproductive success of 
the Northern Goshawk {Accipiter gentilis) . Harvest trials were implemented at 27 of 79 known nest areas, 
and the median post-treatment monitoring period was 3 yr (range = 1-7) . We used a mean nest area 
size of 24 ha, based on the average number and spacing of nests within nest areas, to assess the impact 
of harvesting. Harvesting trials consisted of clearcutting, with the amount of nest area harvested ranging 
from 5-95%. From 1996-2002, we found no significant difference in nest area reoccupation frequencies 
or fledging rates of goshawks between treatment areas and control areas (P > 0.10). Even treatment 
areas with >50% of the nest-area stand removed {N = 7) did not exhibit reduced reoccupation or 
fledging rates. These results are preliminary, pending longer post-treatment monitoring to address high 
annual variation and a potential lag effect that may be exhibited by the goshawks. If these results are 
consistent over a longer period, they may support de-emphasis of management and research effort at 
the nest-area scale and greater emphasis at the territory and landscape scales to examine correlations 
between timber harvesting and territory abandonment and population declines reported in other studies. 

Key Words: Northern Goshawk, Accipiter gentilis; timber harvesting, nest area; reproductive success; adaptive 

management, British Columbia. 


EFECTO DE LA COSECHA DE MADERA CERCA DE LOS SITIOS DE NIDIFICACION SOBRE EL 
EXITO REPRODUCTIVO DE ACCIPITER GENTILIS 

Resumen. — Determinamos el efecto de la cosecha de madera realizada cerca de los nidos sobre el exito 
reproductivo de Accipiter gentilis. Los tratamientos de cosecha fueron implementados en 27 de 79 areas 
de nidificacion conocidas, y la mediana del periodo de observacion post-tratamiento fue de 3 anos 
(rango = 1-7) . Para determinar el impacto de la cosecha de madera, utilizamos un area de nidificacion 
promedio de 24 ha basandonos en el numero y espaciamiento promedio de nidos dentro de las areas 
de nidificacion. Los tratamientos de cosecha consistieron en tala rasa y variaron entre un 5% y un 95% 
de area cosechada del area de nidificacion. Entre 1996 y 2002, no encontramos diferencias significativas 
en las frecuencias de reocupacion de sitios de nidificacion o en las tasas de emplumamiento de los 
halcones entre las areas de los tratamientos y las areas control (P > 0.10). Incluso los tratamientos en 
que se removio >50% del bosque del area de nidificacion {N = 7) no exhibieron tasas reducidas de 
re-ocupacion o de emplumamiento. Estos resultados son preliminares hasta que se obtengan resultados 
de un monitoreo post-tratamiento mas largo para dar cuenta de la alta variabilidad anual y posibles 
efectos retardados que puedan estar exhibiendo los halcones. Si estos resultados son constantes a lo 
largo de un periodo de tiempo mayor, estos pueden apoyar una disminucion del enfasis de los esfuerzos 
de manejo e investigacion a la escala de sitio de nidificacion y un aumento del enfasis a las escalas de 
territorio y de paisaje para examinar las correlaciones entre la cosecha de madera y el abandono de los 
territories y disminuciones poblacionales que se han descrito en otros estudios. 


The Northern Goshawk {Accipiter gentilis) is wide- 
ly recognized as a species sensitive to timber har- 


^ Corresponding author’s email address: wildfor® 

bulkley.net 


vest (Squires and Reynolds 1997). In 1995, British 
Columbia established the Forest Practices Code, 
which strengthened management requirements for 
non-timber resources and included a variety of 
coarse- and fine-filter management strategies for 


335 


336 


Mahon and Doyle 


VoL. 39, No. 3 


wildlife under the Identified Wildlife Management 
Strategy (IWMS; BC Ministry of Environment and 
BC Ministry of Forests 1999). The Northern Gos- 
hawk was identified as a focal species in the fWMS, 
and habitat-management guidelines were devel- 
oped that included protection of nest areas, main- 
tenance of a high proportion of mature and old 
forest in the post-fledging area, and for the threat- 
ened A. g. laingi, broad serai stage targets for the 
foraging area (BC Ministry of Environment and BC 
Ministry of Forests 1999). However, conflicting pol- 
icy limited the number of goshawk territories that 
IWMS guidelines were applied to. For forest man- 
agers, the question became whether alternative 
management strategies that were less conservative 
than the IWMS guidelines could still maintain gos- 
hawk nest-area habitat requirements and reproduc- 
tive success. We have attempted to answer that 
question within an adaptive management frame- 
work by monitoring the response of goshawk re- 
productive success to timber-harvesting trials at 
nest areas. 

Few previous studies have monitored the re- 
sponse of goshawks to timber harvest near occu- 
pied nest areas within an experimental framework. 
Crocker-Bedford (1990) measured the effects of 
timber harvest on goshawk reproduction by ex- 
amining the success of 16-200 ha reserves in main- 
taining goshawk occupation in nest areas sur- 
rounded by large partial-cut units (1000-5000 ha). 
Only 25% of 12 treatment territories were reoc- 
cupied at least once over a 3-yr period, compared 
to 79% of 19 control areas that were reoccupied 
(Crocker-Bedford 1990). Woodbridge and Detrich 
(1994) observed a correlation between nest area 
(nest stand cluster) size and occupancy, with oc- 
cupancy frequencies at stand clusters <20 ha, 40— 
60 ha, and >60 ha of <50%, 75-80%, and nearly 
100%, respectively. In that study, timber harvesting 
was one factor that affected nest-stand cluster size, 
but it was not explicitly isolated from other factors 
potentially affecting stand patterns and sizes. Patla 
(1997) found that occupancy was higher at nest 
areas prior to timber harvesting (79%) than after 
(47%), and that post-harvest areas with >50% oc- 
cupancy had higher percent mature forest cover 
than nest areas with <50% occupancy. Penteriani 
and Faivre (2001) found that goshawk reproduc- 
tive productivity did not differ between shelter- 
wood harvested and untreated nest stands. 

Our study differs from previous work that eval- 
uated the effects of timber management on gos- 


hawk nest area reoccupancy and productivity for 
several reasons: (1) we monitored a larger sample 
of territories than previously studied (27 treatment 
areas and 52 controls); (2) our study was replicated 
in two forest types with ca. equal sample sizes in 
each area; (3) we examined a range of treatment 
levels (amount of nest area removed by clearcut- 
ting), and were able to control those levels exper- 
imentally; and (4) we compared responses pre- and 
post-treatment, as well as post-treatment responses 
to controls. Here we summarize the results of this 
ongoing study from 1996-2002- 

Study Area and Methods 

We replicated this study in two different forest types in 
west-central British Columbia, Canada, with approximate- 
ly equal numbers of nest areas in each. The first study 
area was within the Interior Cedar Hemlock (ICH) and 
Coastal Western Hemlock (CWH) biogeoclimatic zones 
(Banner et al. 1993) in the Kispiox Forest District 
(55°25'N, 127°45'W). This area (ICH/CWH) is along the 
eastern side of the Coast Mountain Range and consists 
of mountain ranges bisected by broad glaciated valleys 
with an elevation range of 200-2500 m. The climate is 
transitional between cool, wet coastal conditions and dri- 
er interior conditions with greater seasonal temperature 
variation. The mean annual precipitation varies from 
600-1200 mm (Banner et al. 1993), with rain occurring 
on half the days during the goshawk breeding seasons we 
monitored. Forests within the ICH and CWH are pre- 
dominantly old growth (>200 yr), coniferous stands 
dominated by western hemlock ( Tsuga heterophylla) , and 
included subalpine fir {Abies lasiocarpa) , western redcedar 
{Thuja plicata) , and Roche spruce {Picea sitchensis X glau- 
ca). Zonal ecosystems consist of hemlock forests with 
moderate-high canopy closure, sparse shrub and herb 
layers, and a thick feathermoss carpet. 

The second study area is 200 km to the southeast in 
the Sub-Boreal Spruce (SBS) biogeoclimatic zone (Ban- 
ner et al. 1993) in the Lakes and Morice Forest Districts 
(N54°25’N, 126°00’W). It occurs on the interior Nechako 
Plateau, with elevations of 500-1000 m. The climate in 
the SBS is primarily continental and is characterized by 
greater seasonal temperature extremes than in the coast 
mountain range, with cold, snowy winters and relatively 
warm, moist, short summers. Annual precipitation is 
440-650 mm (Banner et al. 1993), with rain occurring 
on less than 20% of the days during the breeding seasons 
we monitored. Forests in the SBS have been subject to 
frequent fires (mean fire interval <150 yr), and zonal 
sites are frequently dominated by mature serai stands of 
lodgepole pine {Pinus contorta) with subalpine fir, hybrid 
white spruce {Picea glauca X engelmannii) , and trembling 
aspen {Populus tremuloides) . The shrub and forb layers are 
usually sparse, though variable, and are generally more 
developed than in the ICH. 

In both study areas, ca. 55% of the forested land base 
is mature forest, 25% is young forest, and 20% is in a 
shrub/herb stage. Forestry roads and clearcuts are pres- 
ent in all portions of both study areas, and the latter 


September 2005 


Conservation 


337 


account for the majority of area in the shrub/herb stage. 
Minimum goshawk densities of ca. four pairs per 100 km^ 
are similar between the ICH and SBS based on inventory 
work in core portions of each study area (T. Mahon and 
F. Doyle unpubl. data). Potential avian competitors for 
nest sites and habitat occur at low densities and included 
Red-tailed Hawks {Buteo jamaicensis) , which are found in 
open areas, Barred Owls {Strix varia), mostly in the ICH, 
Great Gray Owls {Strix nebulosa), mostly in the SBS, and 
Great Horned Owls {Bubo virginianus ) , which occur with- 
in riparian and mixed forest habitats at lower elevations 
throughout the region. 

Nest Area Size and Habitat Characteristics. The esti- 
mated size of goshawk nest areas in the literature ranges 
from 8 ha (Reynolds 1983) to 50 ha (McCarthy et al. 
1989). We calculated a theoretical “typical” nest-area size 
in our study based on the mean number of nest sites and 
the mean spacing distance among nest sites for 21 nest 
areas located early in the study and applied a 200 m buff- 
er around the nests. The 200 m buffer was based on ob- 
served distance of nest sites from forest edges, concen- 
trated sign (plucking perches, “white wash” [fecal 
deposits], and roosts), juvenile movements during the 
early post-fledging period, and nest defense behaviors 
displayed by adult birds, which are recognized as key fea- 
tures that determine the boundaries of goshawk nest ar- 
eas (Reynolds et al. 1992, Squires and Reynolds 1997). 
Using our observed mean of three nests per nest area, 
mean spacing of 188 m between nest trees, and a 200 m 
buffer resulted in a nest-area size of 24 ha. 

To test the appropriateness of this theoretical nest area 
size, we overlaid a 24-ha circle on each of the 79 known 
nest areas in 2002 to assess how many nest sites were 
encompassed within the 24-ha circular area. On the basis 
that only 4% of the nest sites fell outside of the 24-ha 
circles, we accepted that this size was the appropriate size 
to use. 

Nest area stands in the ICH/CWH were dominated by 
western hemlock and typically had larger diameter and 
taller trees than in the SBS, which were dominated by 
lodgepole pine, but otherwise habitat characteristics were 
similar between study areas. Most nest areas were in ma- 
ture (>100 yr) or old growth (>240 yr) stands with rel- 
atively closed primary canopies (45-65%) and open sub- 
canopy flyways, on mesic sites. We observed no evidence 
of nest area selection with respect to slope or aspect in 
either study area, except for avoidance of very steep 
slopes (>45%). In most cases, nest areas were located in 
contiguous mature forest matrix, and in all cases suitable 
alternative nest area stands were available within 800 m 
of the original nest area. Forest composition, stand age, 
stand height, and canopy closure did not differ between 
treatment and control nest areas within each study area 
(P> 0.10). 

Experimental Design. We employed an adaptive man- 
agement framework in this study to integrate our re- 
search into operational timber harvesting and to maxi- 
mize the utility of research outcomes to forest managers. 
This approach involved four key steps: (1) defining an 
area ojf scientific uncertainty; (2) developing and imple- 
menting management trials as real world experiments to 
test that uncertainty; (3) evaluating the outcomes of the 
trials; and (4) adjusting management guidelines on the 


basis of the knowledge gained (Morrison et al. 1998) 
The key uncertainty we investigated was how much gos- 
hawk nest area habitat can be removed via clearcutting 
before nest area reoccupation and productivity are im- 
pacted. 

Design of harvesting trials included operational factors 
identified by forest licensees, as well as experimental fac- 
tors associated with our study. In this context, these trials 
were not tightly controlled experiments because we could 
not completely control aspects of the timber harvesting 
relating to pattern and overall size. However, the resul- 
tant harvesting trials do provide a range of scenarios with 
respect to our primary treatment variable (amount of 
nest area harvested). Timber harvesting consisted of 
clearcuts with patch retention. Patch retention areas did 
not have any harvesting within them and were generally 
located to provide a mature forest buffer (25-200 m) 
around known goshawk nest trees. Other mature forest 
patches were occasionally retained in goshawk nest areas, 
including 20-60 m wide riparian buffers and 0.1— 4.0 ha 
upland “wildlife tree patches.” Within the clearcut areas, 
all merchantable trees were removed and in-block reten- 
tion, if any, was limited to sporadic deciduous trees, scat- 
tered advanced regeneration, and occasional snags that 
were topped at 2 m. Timber harvesting was conducted 
outside of the breeding season to minimize the con- 
founding effect of logging disturbance (Toyne 1997). 

We quantified two response variables related to repro- 
ductive success. Our primary variable was the rate of nest 
area reoccupation into the incubation period, which rep- 
resents the evaluation of nest areas by goshawks and their 
commitment to use them. Importantly, we present reoc- 
cupation rates, opposed to occupation rates. This was 
necessary because we found new nest areas each year and 
added them to the study. Therefore, the sample of nest 
areas used to calculate reoccupation rates in year X is 
the sample of nest areas that were known at the end of 
year X — 1. 

We tested for overall differences in reoccupation rates 
between treatments and controls using a chi-square anal- 
ysis and pooled data from study areas and years. To assess 
the effect of treatment level (amount of nest area har- 
vested), we also summarized the reoccupation rates sep- 
arately for treatment areas that had >50% of the nest 
area stand removed and which we had monitored for at 
least 2 yr post-treatment. 

We also examined nest productivity — the number of 
fledglings produced per nesting attempt — as a response 
variable. Nest productivity must be interpreted with cau- 
tion, because once a commitment is made to nest in an 
area, overall fledging rates are more likely dependent on 
breeding season food supply than nest area habitat 
(Doyle 2000). An exception to this would be if timber 
harvesting led to higher nestling depredation rates. To 
address this issue, we evaluated the cause of nestling mor- 
talities whenever possible. Similar to reoccupation rates, 
mean annual fledging rates were summarized excluding 
new nest areas found for that year. We tested for overall 
differences in fledging rates between treatments and con- 
trols using a ^-test, again pooling data from study areas 
and years. 

Nest-tirea Monitoring. We used a combination of telem- 
etry and nest area searches at areas without tagged birds 


338 


Mahon and Doyle 


VoL. 39, No. 3 



0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 60-90 90-100 

Percent of 24-ha Nest Area Removed 

Figure 1. Distribution of treatment levels (amount of 
nest area clearcut) for harvesting trials at 27 Northern 
Goshawk nest areas in west-central British Columbia 
1996-2002. 

to monitor annual reoccupation and fledging rates at 
treatment and control nest areas. Initially, we attempted 
to radio-tag an individual at every treatment area and at 
a subsample of the control areas. However, as the study 
progressed, we determined that nest-area searches were 
sufficient to document reoccupation. Due to the extra 
time and cost associated with radio-tagging, and the po- 
tential negative impacts of radio-tagging to goshawks 
(Reynolds et al. 2004), we reduced our annual sample of 
nest areas with tagged birds to ca. 10% and only tagged 
birds at treatment areas. Adult goshawks were captured 
and tagged during the nestling period and early post- 
fledging period using box traps baited with Rock Pigeons 
{Columba livia; Kenward and Marcstrom 1983) or mist 
nets around a tethered pigeon or owl decoy. Tail-mount- 
ed radios were used instead of backpacks, so that we did 
not have to recapture the birds to remove the tags. 
Tagged birds were monitored the following breeding sea- 
son using ground-based telemetry tracking to determine 
their breeding status and location. 

For nest areas without tagged birds, we conducted in- 


1 5 2 



Year 

Figure 2. Annual reoccupation rates at Northern Gos- 
hawk nest areas at treatment sites, where timber harvest 
occurred, and control areas in west-central British Co- 
lumbia, 1996-2002. Values above the bars equal number 
of nest areas available for reoccupation. 


tensive ground searches within ca. 1 km of the original 
nest area to ascertain the occupancy of each nest area. 
This involved surveying all known nests within a nest 
area, and if none of the known nests were occupied, in- 
tensively searching for new nests and other signs of use 
such as presence of goshawks, “white-wash,” and pluck- 
ing perches. If no occupied nest was found using visual 
searches, we conducted systematic call playback surveys 
(Kennedy and Stahlecker 1993) using a 300 X 300 m grid 
to elicit responses from goshawks in the vicinity. Nest area 
searching was conducted during the courtship, incuba- 
tion, and post-fledging periods. All occupied nests were 
monitored biweekly to determine their success and fledg- 
ing rates. 

Results 

Of the 79 nest areas located in the two study 
areas, harvesting trials were implemented at 27 ar- 
eas (13 ICH/CWH, 14 SBS). The treatment levels 
(amount of nest area clearcut) ranged from 5-95% 
(Fig. 1). The median time since timber harvest at 
treatment nest areas was 3 yr (range = 1-7) . 

We found no difference in reoccupation rates of 
nest areas between treatment and control areas (x^ 
= 0.021, P = 0.89). We combined data from the 
two study areas for analysis because they showed a 
similar pattern of response with reoccupation rates 
for treatments and controls of 54% and 53% in the 
ICH/CWH, and 61% and 63% in the SBS. The 
total reoccupation rates from 1996-2002 were 58% 
at treatment areas (N =73 potential breeding at- 
tempts) and 57% at controls {N = 138; Fig. 2). We 
found consistent patterns of reoccupation rates be- 
tween treatment and controls across years, with 
greater variation among years than between treat- 
ments and controls. 

Seven nest areas had >50% of the nest area 
stand removed. Goshawks returned and bred suc- 
cessfully at all seven of these nest areas in at least 
one year post-treatment. For the years 2000-02 
combined (the post-treatment period for these sev- 
en treatments), the reoccupation rates were 62% 
at treatment areas compared to 50% at controls. 

The mean number of chicks fledged per nesting 
attempt did not differ between treatments (1.63 ± 
1.05 [SD], N= 44) and controls (1.31 ± 1.13, N 
= 73; t — 0.306, P = 0.77). The mean nest pro- 
ductivity by study area was 1.54 ± 0.70 (N = 22) 
for treatments and 1.29 ± 1.03 (N = 35) for con- 
trols in the ICH/CWH and 1.67 ± 1.16 (W = 22) 
for treatments and 1.43 ± 1.06 {N = 38) for con- 
trols in the SBS. 

Discussion 

All nest areas we monitored for ^2 yr showed 
evidence of multiyear use and strong nest-area fi- 


September 2005 


Conservation 


339 


delity. Further, occupancy has been maintained at 
several nest areas where at least one of the original 
occupants has died or disappeared. This is consis- 
tent with other studies that have observed high fi- 
delity even after the nest area is modified (Reyn- 
olds 1983, Woodbridge and Detrich 1994, Patla 
1997). The implication of this behavior is that fi- 
delity to nest areas may override response to re- 
duced suitability and result in a lag effect before 
goshawks relocate to more suitable habitat. 

Another major implication of the nest-area fi- 
delity exhibited by goshawks relates to forest man- 
agement. Effectively, nest-area fidelity is so strong 
in our study areas that nest areas can be consid- 
ered spatially-fixed resources for forest manage- 
ment purposes. Once a nest area is located and 
protected, forest managers can proceed with har- 
vesting in other parts of the territory, as estimated 
by territory spacing data, with low risk of impacting 
another nest area. Where goshawk nest area pro- 
tection is a management objective, this provides 
forest licensees with a strong incentive to locate 
and to maintain nest areas because it reduces po- 
tential constraints within the remainder of the ter- 
ritory. Failing to adequately protect a nest area may 
result in the goshawks relocating to another stand 
scheduled for timber harvesting, which was the 
case with two of three relocations we observed in 
2003. Management strategies to maintain alternate 
nesting habitat, post-fledging area habitat, and suit- 
able foraging habitat would still be desirable at the 
territory scale, but those strategies are typically 
more flexible, at least in a spatial context, than 
protection measures for the nest area. 

In addition to nest-area fidelity, inaccuracy in 
our estimate of nest-area size or variability in sizes 
could also affect the interpretation of our results. 
For example, if our estimates were too large, and 
included area outside of the true nest area, then 
the actual treatment impact would be less than re- 
ported. We estimated the “typical” nest-area size 
in our study areas based on mean number of nest 
sites, spacing between nest sites, evidence of oc- 
cupation, and defensive behavior around nest sites. 
These characteristics were variable among nest ar- 
eas, which probably corresponded to different 
nest-area sizes. We considered estimating the size 
of each nest area individually, but decided that 
would be even more problematic and biased than 
our systematic approach. Because of the uncertain- 
ty associated with nest-area size and its relationship 
to estimated treatment level, we did not focus our 


analysis on treatment level beyond two classes: all 
treatment areas and nest areas where >50% of the 
stand have been clearcut. 

Despite nest-area fidelity by goshawks and poten- 
tial lag effects, our study supports the findings of 
other research (Penteriani and Faivre 2001) that 
indicated goshawks tolerated modification of nest 
area stands, or relocated to new stands, without de- 
creased reproductive output (assuming that alter- 
native nest area habitat was available and distrib- 
uted within the landscape appropriate to goshawk 
territory spacing). In Italy and France, Penteriani 
and Faivre (2001) reported a similar response by 
goshawks to shelterwood harvesting. They found 
that breeding frequency and the number of young 
produced per breeding pair did not differ between 
logged and unlogged stands. They also reported 
that where timber harvesting exceeded 30% of the 
nest stand, goshawks often relocated to the neigh- 
boring mature stands, but that overall reproductive 
success was not affected. 

Several independent studies in Fennoscandia 
have shown that goshawk populations declined by 
50-60% from the 1950-80s (Widen 1997). Widen 
(1997) examined several factors most often asso- 
ciated with declining raptor populations, including 
pesticides, persecution, prey populations, and nest- 
ing and foraging habitat loss associated with forest 
development. Of these factors, only decreases in 
the amount and patch size of mature forest at the 
foraging habitat scale showed a clear correlation 
with the decline in goshawk populations. 

North American studies that suggested de- 
creased nesting productivity in response to timber 
harvesting (Crocker-Bedford 1990, Patla 1997) 
were limited by their study design or by issues re- 
garding scale of analysis relative to scale of har- 
vesting. In Idaho, Patla (1997) compared repro- 
ductive success pre- and post-treatment, but not 
post-treatment areas to controls. Pre- and post- 
treatment comparisons in the absence of controls 
depend on the assumption that other factors af- 
fecting reproductive success are similar over the 
entire monitoring period, or at least have a minor 
effect relative to the treatment effect being stud- 
ied. However, the reproductive success of goshawks 
is known to vary considerably from year to year 
depending on prey abundances (Doyle and Smith 
1994) and weather (Younk and Bechard 1994). 

Crocker-Bedford (1990) examined 16-200 ha 
nest area reserves surrounded by large partial cuts 
in Arizona and found much lower occupancy in 


340 


Mahon and Doyle 


VoL. 39, No. 3 


the logged territories than at controls. However, 
the timber harvesting being evaluated was carried 
out over 1000-5000 ha units, which would have in- 
fluenced both nesting and foraging area suitability. 
Crocker-Bedford (1995) later reanalyzed reoccu- 
pation rates and nestling production with respect 
to amount of harvesting that had occurred at the 
home-range scale of 2290 ha and found an inverse 
correlation between the reproductive success vari- 
ables and harvesting. 

To address high annual variation and a potential 
lag effect in responses by the goshawks, we will con- 
tinue this study through 2005. If our longer-term 
results are consistent with Penteriani and Faivre 
(2001) and continue to show no decreased repro- 
ductive success by goshawks at nest areas modified 
by timber harvesting, it would support Widen ’s 
(1997) theory that habitat changes at the foraging 
area scale are the primary factor affecting goshawk 
populations. Notwithstanding the need for addi- 
tional manipulative studies at the nest-area scale, 
we recommend that research on goshawks needs 
to shift from descriptive nest-area scale studies, 
which are numerous, to territory and landscape 
scale studies, which are few. Specifically, research 
should attempt to examine habitat requirements at 
the territory and landscape scale that can be in- 
corporated into forest management strategies, 
such as serai-stage and patch-size distributions. 

Acknowledgments 

This project was possible due to the contributions and 
collaborations of numerous biologists, foresters, and 
planners from multiple forest licensees and government 
agencies within British Columbia’s forest sector. We ac- 
knowledge the participation of Babine Forest Products, 
Houston Forest Products, Skeena Cellulose Inc., BC Min- 
istry of Forests, BC Timber Sales, BC Ministry of Sustain- 
able Resource Management, and BC Ministry of Water, 
Land, and Air Protection. Melissa Todd was instrumental 
in initiating goshawk research in our study areas. We also 
thank Don Reid, Karen Grainger, Norm Bilodeau, and 
Anne Hetherington for their contributions. For field as- 
sistance, we thank Rob Kelly and Mike Nelligan. Our re- 
search was funded by the BC government through the 
Forest Renewal BC program and the Forest Investment 
Account and by BC Timber Sales, Babine Forest Prod- 
ucts, Houston Forest Products, and Skeena Cellulose Inc. 
This manuscript was improved by comments from C.L. 
Mahon and two anonymous reviewers. 

Literature Cited 

Banner, A., W. Mackenzie, S. Haeussler, S. Thompson, 
J. PojAR, AND R. Trowbridge. 1993. A field guide to 
site identification and interpretation for the Prince 


Rupert Forest Region. Ministry of Forests, Research 
Branch, Victoria, BC, Canada. 

BC Ministry of Environment, Lands and Parks and BC 
Ministry of Forests. 1999. Managing Identified Wild- 
life Strategy. Forest Practices Code of British Colum- 
bia. Forest Practices Code of British Columbia. Vic- 
toria, BC, Canada. 

Crocker-Bedford, D.C. 1990. Goshawk reproduction 
and forest management. Wildl. Soc. Bull. 18:262-269. 

. 1995. Northern Goshawk reproduction relative 

to selection harvest in Arizona./. Raptor Res. 29:42. 

Doyle, F.I. 2000. Timing of reproduction by Red-tailed 
Hawks, Northern Goshawks, and Great Horned Owls 
in the Kluane Boreal Forest of Southwestern Yukon. 
M.S. thesis, Univ. British Columbia, Vancouver, BC, 
Canada. 

and J.M.N. Smith. 1994. Population responses of 

Northern Goshawk to the 10-year cycle in numbers of 
snowshoe hares. Stud. Avian Biol. 16:122-129. 

Fairhurst, G. and M. Bechard. 2005. Relationships be- 
tween winter and spring weather and Northern Gos- 
hawk {Accipiter gentilis) reproduction in northern Ne- 
vada. J. Raptor Res. 39:229-236. 

Kennedy, P.L. and D.W. Stahlecker. 1993. Responsive- 
ness of nesting Northern Goshawks to taped broad- 
casts of 3 conspecific calls. J. Wildl. Manag. 57:249- 
257. 

Kenward, R.E. AND V. Marcstrom. 1983. The price of 
success in goshawk trapping./. Raptor Res. 17:84—91. 

McCarthy, C., W.D. Carrier, and W.F. Laudenslayer. 
1989. Coordinating timber management activities 
with raptor nesting habitat requirements. Pages 229- 
235 in B.G. Pendleton, C.E. Ruibol, P.L. Krahe, K 
Steenhof, M.N. Kochert, and M.N. LeFranc,Jr. [Eds.], 
Proceedings of the western raptor management sym- 
posium and workshop. National Wildlife Federation, 
Washington, DC U.S.A. 

Morrison, M.L., B.G. Marcot, and R.W. Mannan. 1998. 
Wildlife habitat relationships: concepts and applica- 
tions. Univ. Wisconsin Press, Madison Wl U.S.A. 

Patla, S.M. 1997. Nesting ecology and habitat of the 
Northern Goshawk in undisturbed and timber harvest 
areas on the Targhee National Forest, greater Yellow- 
stone ecosystem. M.S. thesis, Idaho State Univ., Boise, 
ID U.S.A. 

Penteriani, V. and B. Fatvre. 2001. Effects of harvesting 
timber stands on goshawk nesting in two European 
areas. Biol. Conserv. 101:211—216. 

Reynolds, R.T. 1983. Management of western coniferous 
forest habitat for nesting accipiter hawks. USDA For- 
est Service, Fort Collins, CO U.S.A. 

, R.T. Graham, M.H. Reiser, R.L. Basset, P.L. Ken- 
nedy, D.A. Boyce Jr., G. Goodwin, R. Smith, and E.L. 
Fisher. 1992. Management recommendations for the 
Northern Goshawk in the southwestern United States. 
USDA Forest Service, Fort Collins, CO U.S.A. 

, G.C. White, S.M. Joy, and R.W. Mannan. 2004. 


September 2005 


Conservation 


341 


Effects of radiotransmitters on Northern Goshawks: 
do tailmounts lower survival of breeding males? J. 
Wildl. Manag. 68:25-32. 

Squires, J.R. and R.T. Reynolds. 1997. Northern Gos- 
hawk. In A. Poole and F. Gill [Eds.], The birds of 
North America, No. 298. The Birds of North America, 
Inc., Washington, DC U.S.A. 

Toyne E.P. 1997. Nesting chronology of Northern Gos- 
hawks, {Accipiter gentilis) in Wales: implications for for- 
est management. Forestry 70:121-127. 

Widen, P. 1997. How and why, is the goshawk {Accipiter 


gentilis) affected by modern forest management m 
Fennoscandia? J. Raptor Res. 31:107-113. 

Woodbridge, B. and P.J. Detrich. 1994, Territory occu- 
pancy and habitat patch size of Northern Goshawks 
in the southern Cascades of California. Stud. Avian 
Biol. 16:83-87. 

Younk, J.V. AND M.J. Bechard. 1994. Breeding ecology of 
the Northern Goshawk in high-elevational aspen for- 
ests of northern Nevada. Stud. Avian Biol. 16:119—121. 

Received 4 February 2004; accepted 14 February 2005 

Associate Editor: Michael I. Goldstein 


/. Raptor Res. 39(3) :342-350 
© 2005 The Raptor Research Foundation, Inc. 


A REVIEW OF THE STATUS AND DISTRIBUTION OF NORTHERN 

GOSHAWKS IN NEW ENGLAND 

Stephen DeStefano^ 

U.S. Geological Survey, Massachusetts Cooperative Fish and Wildlife Research Unit, 

Holdsworth Natural Resources Center, University of Massachusetts, Amherst, MA 01003 U.S. A. 


Abstract. — The Northern Goshawk {Accipiter gentilis) is a resident breeder throughout much of the 
forested landscape of New England and a winter resident in most of New England, except possibly for 
extreme northern portions. Historically, goshawk numbers and distribution presumably declined as 
agriculture and logging grew to dominate the region in the 19* century when large parts of New 
England were cleared upwards of 75% of the forest cover. Goshawks likely responded to reforestation 
during the middle and latter decades of the 20* century. However, most biologists agree that although 
goshawk numbers may be stable or perhaps increasing slightly today, their true status and distribution 
in this six-state region is largely unknown. Goshawks in New England nest in mature regrown coniferous, 
deciduous, and mixed forest. From a landscape perspective, conservation, maintenance, and enhance- 
ment of mature forest, as well as early successional-stage cover, are both necessary for this species in 
New England. Restoration and management of these cover types would benefit not only goshawks and 
their prey, but also a significant portion of the region’s biodiversity. Because of the extensive and inten- 
sive relationships humans have had with the New England landscape over the past three centuries, the 
region would make a valuable subject area for long-term monitoring and research on a wide-ranging 
top-level predator such as the goshawk. 

Key Words: Northern Goshawk, Accipiter gentilis; Connecticut, distribution; Maine; Massachusetts; New 

Hampshire, New England; northeastern U.S.; Rhode Island; Vermont, status. 


UNA REVISION SOBRE EL ESTADO Y LA DISTRIBUCION DE ACCIPITER GENTILIS EN NUEVA 
INGLATERRA 

Resumen. — Accipiter gentilis es un ave residente que nidifica a lo largo de la mayor parte de los bosques 
de la region de Nueva Inglaterra, y un residente invernal en casi toda Nueva Inglaterra con excepcion 
posiblemente de las porciones mas extremas del norte. Historicamente, los numeros y la distribucion 
de A. gentilis presumiblemente disminuyeron a medida que la agricultura y la tala aumentaron hasta 
dominar la region durante el siglo XIX, cuando grandes partes de Nueva Inglaterra fueron deforestadas, 
transformandose mas del 75% de la cobertura del bosque. Luego A. gentilis probablemente respondio 
a la reforestacion a partir de mediados del siglo XX. Sin embrago, la mayoria de los biologos coinciden 
con que, aunque los numeros de A. gentilis pueden permanecer estables o tal vez haber incrementando 
levemente en la actualidad, su verdadero estatus y distribucion son basicamente desconocidos en esta 
region que comprende seis estados. A. gentilis nidifica en Nueva Inglaterra en bosques regenerados 
maduros de coniferas, en bosques deciduos y en bosques mixtos. Desde una perspectiva del pais^e, la 
conservacion, mantenimiento y mejoramiento del bosque maduro y de las etapas sucesionales tempran- 
as, son una preocupacion en Nueva Inglaterra. La restauracion y el manejo de estos tipos de cobertura 
beneficiarian no solo a A. gentilis y a sus presas, sino tambien a una porcion significativa de la biodiv- 
ersidad de la region. Debido a las relaciones extensas e intensas que los humanos han tenido con el 
paisaje de Nueva Inglaterra a lo largo de los ultimos tres siglos, la region seria un area piloto interesante 
y valiosa para el monitoreo y la investigacion a largo plazo de un depredador tope con un area de 
accion amplia como A. gentilis. 

[Traduccion del equipo editorial] 


^ Email address: sdestef@forwild.umass.edu 


342 


September 2005 


Conservation 


343 


Much attention has focused on the Northern 
Goshawk {Accipiter gentilis) in the western United 
States west of the 100* meridian (Kennedy 1997, 
Crocker-Bedford 1998, DeStefano 1998, Andersen 
et al. 2003). However, the species is holarctic in 
distribution and is found in boreal and northern 
temperate forests in the northern hemisphere of 
North America and Eurasia (Squires and Reynolds 
1997). In the northeastern U.S., the goshawk is 
found regularly throughout this region, including 
in all six New England states and as far south as 
Maryland and West Virginia (Squires and Reynolds 
1997). 

Much of the interest and concern for goshawks 
in the western U.S. is related to forest management 
practices, in particular the cutting of large trees 
and conversion of the forested landscape from late 
to early-seral-stage forest (DeStefano 1998). How- 
ever, in the eastern U.S., woody vegetation and re- 
growth forest has increased to such an extent that 
biologists are now concerned with the lack of early- 
seral-stage habitats, such as grasslands and shrub- 
lands, and the loss of some forest types such as 
aspen {Populus spp.) and the species they support- 
ed (Askins 2001, Thompson and DeGraaf 2001). 

The northeastern U.S., and New England in par- 
ticular, have a long history of human occupation 
and land-use change, even before European settle- 
ment (Cronin 1983). In the 18* and 19* centuries, 
clearing for agriculture and timber altered the en- 
tire region (DeGraaf and Yamasaki 2001). Much of 
New England is reforested today, and it is unknown 
but unlikely that these second- or multiple-growth 
forests are similar — and certainly are not identi- 
cal — to the original forests of 300-350 yr ago (Cog- 
bill et al. 2002). 

The primary question in the eastern U.S., one 
that has implications for goshawk management in 
the western U.S., is what is the status and distri- 
bution of Northern Goshawks in the greatly trans- 
formed landscapes of the Northeast? The objec- 
tives of this paper are to examine that question by 
reviewing recent accounts and expert opinion on 
the status of Northern Goshawks, describe the dis- 
tribution of goshawks in the New England states in 
light of historical changes and current conditions, 
and attempt to assess the status of the species in 
this region. I then make suggestions for potential 
long-term, multi-state research over large land- 
scapes in New England. 


Study Area 

I restricted my review of the status and distribution of 
Northern Goshawks in the Northeast to New England. 
The six New England states (Connecticut, Rhode Island, 
Massachusetts, Vermont, New Hampshire, and Maine) 
cover ca. 163 200 km^ and form an identifiable political 
and regional entity. This review could have also included 
New York, Pennsylvania, Maryland, New Jersey, and other 
states to the south, but New England forms a convenient 
and manageable region for addressing questions of status 
and distribution. More importantly, there are more ex- 
tensive long-term records and documentation of land-use 
change for New England than any other region of the 
country (Cogbill et al. 2002, Foster 2002), in addition to 
well-studied, species-habitat relationships (DeGraaf and 
Yamasaki 2001). Nonetheless, many other parts of the 
Northeast share similar land-use histories with New En- 
gland, and at least some of the insights and speculation 
provided here regarding the status and distribution of 
Northern Goshawks in New England will be similar for 
other northeastern states. 

New England is diverse in vegetation, topography, cli- 
mate, and other ecological factors, but in general is dom- 
inated by deciduous, mixed deciduous-coniferous, and 
coniferous forest as one moves from south to north 
(DeGraaf and Yamasaki 2001). Summers are warm and 
humid; winters are usually cold and snowy. Precipitation 
in the form of rain and snow is highly variable and based 
on many factors, such as latitude, elevation, and prox- 
imity to the coast, but generally ranges from 90-140 cm 
annually. Numerous lakes, ponds, rivers, and wetlands 
cover the region. Major mountain ranges include the 
Berkshire Mountains, which extend from western Con- 
necticut through Massachusetts, the Taconic Mountains 
of Massachusetts, Green Mountains of Vermont, and 
White Mountains and Mahoosics of New Hampshire and 
Maine. The entire region was glaciated, and erosion has 
been a major influence on the landforms present today. 

Six forest regions have been identified in New En- 
gland. Major tree species that characterize some of these 
regions include pitch pine {Pinus rigida), oaks {Quercus 
spp.), eastern hemlock (Tsuga canadensis), eastern white 
pine {Pinus strobus) , red spruce {Picea rubens) , and balsam 
fir {Abies balsamea). These forest regions, in a general 
south to north distribution, are pitch pine-oak (on Cape 
God), central hardwoods-hemlock-white pine, transition 
hardwoods-white pine, northern hardwoods, northern 
hardwoods-spruce, and spruce-fir (DeGraaf and Yamasaki 
2001). American beech {Fagus grandifolia) , birches {Bet- 
ula spp.), sugar maple {Acer saccharum) and several other 
maples, hickories {Carya spp.), ashes {Fraxinus spp.), 
cherries {Prunus spp.), and aspens are other major tree 
species. Disturbance to forest growth and structure is 
common in New England; DeGraaf and Yamasaki (2001) 
identify and discuss five major types of disturbances that 
have altered New England’s forest, including windthrow, 
fire, exotic pests and pathogens, agriculture, and logging. 

Much of southern New England is highly urbanized, 
with some of the highest densities of people in the coun- 
try. However, substantial portions of the region are still 
rural. Most of the forest land (some 120 000 km^ or 
>70%) is privately owned by >760 000 different owners 
and divided into small parcels and woodlots of nonin- 


344 


DeStefano 


VoL. 39, No. 3 


dustrial-private forest (commonly abbreviated as NIPF 
lands), but there are large privately owned commercial 
timberlands in the north, particularly in Maine (Birch 
1996). Federal land is much less common in the East 
than the West, but there are two national forests in the 
region: the Green Mountain National Forest in Vermont 
and the White Mountain National Forest in New Hamp- 
shire and Maine. 

Methods 

I reviewed written accounts, both recent and historical, 
of the Northern Goshawk and related land-use changes 
in New England. I also summarized information reported 
from breeding bird atlases, which have been published 
for all six states. I examined long-term trends in numbers 
for both breeding and wintering goshawks by querying 
web databases for the North American Breeding Bird 
Survey (BBS; http://www.mp2-pwrc.usgs.gov/bbs/) and 
the National Audubon Society’s Christmas Bird Counts 
(CBC; http://audubon2.org/birds/cbc/hr/graph.html) . 
Finally, I queried local experts in each state to gather 
their knowledge on the status and distribution of gos- 
hawks. I defined an expert as anyone currently working 
as a professional biologist with a state or federal agency 
or a recognized non-governmental organization, who 
had a focus on raptors, threatened or endangered spe- 
cies, or forest wildlife. I asked a series of eight questions, 
which addressed issues related to status, distribution, 
population trends, habitat use, relationship to mature 
forest and young, early-successional forest, and prey. 
Questions were sent to biologists in each state and at the 
Green Mountain and White Mountain National Forests. 

RESUI.TS 

Historical and Recent Accounts of New England 
Forests. Before European settlement, the North- 
east was probably a mix of forested and open hab- 
itats. Native prairie and forests cleared by Native 
American activities were common in southern New 
England, while beaver ( Castor canadensis) meadows, 
periodic fires, and hurricanes created a shifting 
mosaic of forest and open habitats throughout the 
region (Cronin 1983, DeGraaf and Yamasaki 2001, 
Lorimer 2001, Parshall and Foster 2002). Interior 
and northern regions were more heavily forested 
than coastal sections or lands along major rivers 
(DeGraaf and Yamasaki 2001). 

The history of New England since the time of 
European settlement embodies major and constant 
anthropogenic change (Hall et al. 2002) . Foster et 
al. (2002) characterized these changes to the New 
England landscape as a continual transformation 
involving deforestation, intensive agriculture, farm 
abandonment, reforestation, and human popula- 
tion increase. Land was first cleared slowly for set- 
tlements and agriculture until the 1750s, after 
which the pace accelerated until 75% of the arable 


land in central and southern New England was in 
pasture and crops by the first half of the 1800s 
(DeGraaf and Yamasaki 2001). Many of the largest 
trees, such as eastern white pines, were cut to pro- 
vide masts for ships, first for the British navy before 
the Revolutionary War and then for the U.S. navy 
after the war (Walker 1999). Around 1910, the last 
major logging occurred when primarily white 
pines were harvested. These sites grew into hard- 
woods and supported large populations of Ruffed 
Grouse {Bonasa umbellus) during the 1920s and 
1930s (DeGraaf and Yamasaki 2001). Today, about 
65% of southern New England and >90% of 
northern New England are forested (DeGraaf and 
Yamasaki 2001). Each year the age and extent of 
forest in southern and central New England in- 
creases (Brooks and Birch 1988, DeGraaf and Ya- 
masaki 2001). 

Today, the evolution of the New England land- 
scape is marked at least partially by what is no lon- 
ger there. Remnants of what may be called old- 
growth forest make up <1% of the forests of New 
England (Davis 1996, Cogbill et al. 2002). Thus, 
the woodlands of the Northeast could be described 
as multiply-regrown forests of medium-sized and 
medium- to mature-age (40-100 yr) trees. Old- 
growth or virgin forest remnants remain in small 
and scattered amounts, but are essentially ecolog- 
ically extinct, while open grasslands, shrubby hab- 
itats, or young invasive forest types have given way 
to altered disturbance regimes and woody plant 
succession (Lorimer 2001). A dominant canopy 
and major mast-producing tree species, the Amer- 
ican chestnut ( Castanea dentata) , was eliminated as 
a canopy tree by the chestnut blight ( Cryphonectna 
parasitica), introduced from Europe in the early 
1900s (Paillet 2002). American chestnuts still exist 
in the woodlands of New England and elsewhere, 
but never achieve maturity and survive today only 
in the form of sprouts originating from trees or 
seedlings that were established before the arrival 
of the blight (Paillet 2002). The hemlock woolly 
adelgid {Adelges tsugae), an aphid-like insect from 
Japan, has already caused the loss of large numbers 
of eastern hemlock trees in southern New Eng- 
land, and is migrating north, threatening the ex- 
istence of this long-lived, shade tolerant species 
(Orwig et al. 2002) . Changes due to direct mortal- 
ity as well as increased logging, which is occurring 
at a greater rate because of the threat of the loss 
of trees, have led to thinning canopies (Kizlinski et 


September 2005 


Conservation 


345 


al. 2002) and changes in avian communities (Ting- 
ley et al. 2002). 

Many large mammals, such as elk {Cervus ela- 
phus) and caribou {Rangifer tarandus), have been 
extirpated, as have some major predators, such as 
wolves {Canis lupus) and mountain lions {Puma 
concolor, DeGraaf and Yamasaki 2001). Wolves were 
extirpated around 1900, and soon afterwards coy- 
otes {Canis latrans) began colonizing the region 
from the Midwest (Parker 1995). Passenger Pi- 
geons {Ectopistes migratorius) , whose numbers quite 
possibly ranged in the billions and were likely a 
major prey item for goshawks and other raptors, 
went extinct at the turn of the 19* century (Block- 
stein 2002) . By shear numbers alone, their impact 
was a major driving force on the characteristics of 
eastern forests (Ellsworth and McComb 2003) . 
During this recent history, other wildlife species 
have either increased their range or have become 
more common, such as moose {Alces alces) , beaver, 
coyote, fisher {Maries pennanti), Wild Turkey {Me- 
leagris gallopavo), Mourning Dove {Zenaida ma- 
crouro), and others (DeGraaf and Yamasaki 2001: 
13). 

Historical Accounts for the Northern Goshawk. 

In New England, nesting habitat of Northern Gos- 
hawks decreased as forests were cleared for settle- 
ment and agriculture (Bent 1937, DeGraaf and Ya- 
masaki 2001). This was an obvious change in 
habitat for the goshawk, but equally important may 
have been the extinction of the Passenger Pigeon, 
which was likely important prey for goshawks (Bent 
1937). Thus, the Northern Goshawk may have 
been a rare nesting species in New England at the 
turn of the 19* to the 20* century (Bevier 1994). 
It was called a casual species in summer (Forbush 
1925-29), very rare (Bagg and Eliot 1937), and a 
rare and irregular winter resident (Sage et al. 
1913). 

In the late 1800s and early 1900s, the goshawk 
was a rare summer resident in northern New 
Hampshire (Allen 1903, Hoffman 1904, Foss 1994) 
and was seen in southern New Hampshire primar- 
ily as a winter visitor (Dearborn 1903, Foss 1994). 
The discovery (or rediscovery) of the first goshawk 
nest in Massachusetts has been attributed to two 
officials of the Harvard Forest in Petersham, cen- 
tral Massachusetts, in 1922-23 (Wetherbee 1945). 
In some winters, goshawks were reported to come 
out of the north in great numbers to “wreak havoc 
with the grouse of the county” (Wetherbee 1945: 
38). Over 20 skins were reported collected from 


1883-1935 (Wetherbee 1945:117-18). In 1945, 
Wetherbee (1945:23) reported that the “eastern” 
goshawk was among several species of birds that 
have “nested in the past but have doubtful nesting 
status at present.” 

Since about 1955, however, there is some evi- 
dence that both numbers of nesting pairs and the 
range of breeding goshawks have increased steadily 
in New England (DeGraaf and Yamasaki 2001) . For 
example, only three nesting records existed in Ver- 
mont before 1933, but now goshawks nest through- 
out northern New England (Laughlin and Kibbe 
1985, DeGraaf and Yamasaki 2001). DeGraaf and 
Yamasaki (2001) attribute range expansion and an 
increase in population size to the regrowth of New 
England forests. 

Recent Accounts for the Northern Goshawk. 

DeGraaf and Yamasaki (2001) list the Northern 
Goshawk as uncommon to rare, but increasing, in 
New England. They state that goshawks breed 
throughout the New England states and winter 
throughout the region, except for northernmost 
Maine. The Northern Goshawk was one of 41 
breeding bird species that DeGraaf and Yamasaki 
(2001) listed as having “increased significantly in 
abundance” in Massachusetts (Veit and Petersen 
1993). 

State Accounts. The following accounts for each 
of the six New England states were excerpted from 
the atlas of breeding birds for each state and other 
sources as cited. 

Connecticut. Bevier (1994) described the goshawk 
as an uncommon permanent resident and mi- 
grant. Nesting concentrated in higher elevations of 
western Connecticut, where pairs usually occupy a 
territory throughout the year. They exhibit “flexi- 
ble habitat selection,” nesting in tracts of mixed 
northern hardwoods and conifers, especially east- 
ern hemlock and white pine, pure stands of ma- 
ture white or red pine {Pinus resinosa) within more 
extensive tracts of deciduous woods, wetlands, and 
second-growth, deciduous stands. Nesting occurs 
on hillsides, frequently near wetlands and away 
from human disturbance. Prey brought to nests 
was mostly squirrels and chipmunks, grouse, song- 
birds, and waterfowl. 

Rhode Island. Enser (1992) reported that histor- 
ical nesting was unknown. Northern Goshawks may 
now be the most common nesting Accipiter in 
Rhode Island, but there are still very few known 
nests (ca. eight confirmed or possible occurrences 
in the early 1990s). This species became reestab- 


346 


DeStefano 


VoL. 39, No. 3 


lished in the mid-1950s. They usually breed in iso- 
lated areas of coniferous forest, particularly mature 
stands of hemlock and white pine and also decid- 
uous woodlots. 

Massachusetts. Veit and Petersen (1993) listed the 
goshawk as one of 41 breeding bird species whose 
numbers have increased significantly since the 
1950s, based on Griscom and Snyder’s (1955) ac- 
counts. The current status is given as an uncom- 
mon resident and migrant on the mainland and a 
rare migrant on the islands of Nantucket and Mar- 
tha’s Vineyard. In 1995, nesting was restricted to 
western Massachusetts, but now occurs regularly 
throughout the state, except for Cape Cod and the 
Islands. Goshawk numbers fluctuate annually, but 
have been increasing steadily since the mid-1950s, 
both during the breeding season and in winter. 

Vermont. Laughlin and Kibbe (1985) reported 
that goshawks were found almost statewide, but 
were largely confined to areas with medium to 
high relief (e.g., in the Champlain Lowlands along 
Lake Champlain). All but one record were from 
the hilly eastern and southern portions of that re- 
gion. 

New Hampshire. Foss (1994) described the gos- 
hawk as much more common in southern New 
Hampshire in recent decades, while the Cooper’s 
(Accipiter cooperii) and Sharp-shinned {A. striatus) 
hawks seem to have made only modest recoveries 
since the use of DDT was banned in the early 
1970s. The goshawk breeds throughout the state, 
typically in higher elevations, and often nests in 
deciduous trees, especially white birch {Betula pa- 
pyrifera) , red maple {Acer rubrum) , and black birch 
{B. lenta), but occasionally in white pine. Prey 
items include grouse, crows, waterfowl, small birds, 
hares, squirrels, and chipmunks. 

Maine. Adamus (1987) reported that the gos- 
hawk was somewhat common in the central and 
southern parts of the state, but less so further 
north. Confirmed nesting records exist for coastal 
regions and southern and central Maine. Probable 
breeding records exist throughout Maine, includ- 
ing the north-central region and along the Cana- 
dian border. Goshawks are generally absent in 
northern Maine during winter (DeCraaf and Ya- 
masaki 2001). 

Breeding Bird Surveys and Christmas Bird 
Counts. Both breeding (from BBS data) and winter 
(from CBC data) distribution maps show the gos- 
hawk present throughout all of New England, but 
in both cases the number of observations of indi- 



Fig. 1. During the Christmas Bird Count in New Eng- 
land from 1959-60 to 2002-03, counts of Northern Gos- 
hawks have shown a long-term increasing trend. Data 
compiled from National Audubon Society, Inc. web site 
Christmas Bird Count home page (http:/ /www. 
audubon.org/bird/cbc/) for Connecticut, Rhode Island, 
Massachusetts, Vermont, New Hampshire, and Maine. 

vidual birds is <1 per route (Sauer et al. 2003). In 
spite of extremely low densities, long-term CBC 
data show a concurrent long-term increase in sight- 
ings of goshawks, but a slight decrease in number 
of birds observed per unit effort (Fig. 1 ) . 

Expert Opinions. Several biologists responded 
to my questions about goshawks in their state. Not 
unexpectedly, the distribution of Northern Gos- 
hawks in most New England states is somewhat eas- 
ier to determine, and thus better known, than pop- 
ulation status or trends. Breeding bird surveys 
probably best indicate the distribution of nesting 
pairs. In general, goshawks can be found in forest- 
ed areas throughout New England, although den- 
sities could be expected to vary among regions (C. 
Gaughan, S. Melvin, S. Parren, and T. Hodgman 
pers. comm.). In short, most biologists described 
the goshawk as uncommon but present, and given 
naturally low densities of this species, well distrib- 
uted in forested habitat. 

State biologists recognize that information on 
population trends is lacking. Some have stated 
that, although it is commonly reported that gos- 
hawk numbers may be increasing because of wide- 
spread reforestation, there are no definitive data 
to support this proposal. Goshawk numbers may 
have decreased in northern Maine during the 
1960s through 1980s because of widespread spruce 
budworm ( Choristoneura fumiferana) infestations 
and subsequent increased tree mortality and sal- 
vage harvests; however, numbers there may have 
stabilized in the last decade (T. Hodgman pers. 


September 2005 


Conservation 


347 


comm.). It appears that goshawks have expanded 
south in New Hampshire, suggesting that numbers 
have increased in the southern part of the state in 
recent years (C. Gaughan pers. comm.). 

Biologists from the Green and White Mountain 
National Forests provided responses similar to 
those of state biologists regarding the status and 
distribution of Northern Goshawks on their areas 
(C. Grove and M. Yamasaki pers. comm.). Gos- 
hawks are distributed throughout most or all of 
both Vermont’s Green Mountain and New Hamp- 
shire and Maine’s White Mountain National For- 
ests, certainly as breeders and probably as winter 
residents, although some birds may be winter mi- 
grants from the north. Goshawks are not common, 
but neither are they considered rare; the term “un- 
common breeder’’ might best describe their status 
on national forest lands within New England. For- 
est biologists believe that goshawk numbers are 
probably stable at some undetermined level, and 
may even be increasing as suggested by state breed- 
ing bird atlas accounts, but again caution that data 
are lacking and opinions on population trends are 
speculative. On both the Green and White Moun- 
tain National Forests, goshawks nest in mature 
stands of white pine or mixed spruce-fir and hard- 
woods. Given the land-use history of New England, 
many of these stands are essentially regrown ma- 
ture forest of 80-100 yr. Often there are forest 
openings, such as roads, trails, and upland open- 
ings nearby, but usually nests are away from high 
levels of human activity. Some additional general- 
izations of nest sites include gentler slopes at lower 
elevations (e.g., below 450 m). The Northern Gos- 
hawk was listed as a Regional Forester Sensitive 
Species in 2003 on some national forests in the 
northern portions of the U.S. Forest Service’s Re- 
gion 9, but not on either the Green or White 
Mountain National Forest (M. Yamasaki pers. 
comm.). 

Discussion 

Foster et al. (2002) characterized six major tra- 
jectories of change in the long-term dynamics of 
wildlife populations in the northeast; (1) many 
large mammals and birds that declined historically 
have increased recently, (2) open-land species went 
from low to high abundance wdth land clearing, 
but are in decline today, (3) some species were ex- 
tirpated, (4) some species have expanded their 
ranges into the northeast, (5) introduced non-na- 
tives have proliferated, and (6) some persistent 


species did not exhibit major long-term trends. 
Likewise, DeGraaf and Yamasaki (2001) identified 
three major trends in New England’s wildlife in the 
last several decades: (1) forest species are increas- 
ing, (2) grassland and shrubland species are de- 
clining, and (3) many southern birds are spreading 
northward into the region. In addition, a few spe- 
cies like Common Ravens (Corvus corax) and 
moose have extended their range southward. 

At least some of these statements apply directly 
or indirectly to the Northern Goshawk in New En- 
gland. The goshawk was apparently one of those 
forest species that has increased in numbers in the 
last half century. This was probable given that at 
least three quarters of New England’s forests were 
cleared for agriculture and high-graded for timber. 
The number and distribution of goshawks could 
have been expected to decline significantly with 
the amount of forest clearing that occurred in the 
18* and 19* centuries. With reforestation occur- 
ring during the middle decades of the 20* century, 
the distribution and number of goshawks likely in- 
creased. This presumed long-term decline followed 
by an increase in numbers of goshawks must be 
viewed in the proper temporal scale: in decades, if 
not centuries, of change. More difficult to deci- 
pher is whether or not goshawk distribution and 
numbers are increasing today. Some evidence in- 
dicates that this is the case, but empirical data are 
extremely limited to nonexistent. Thus, it is diffi- 
cult to speculate on recent (say, the last 20-30 yr) 
population trends without more definitive data. 
However, long-term efforts, such as the Christmas 
Bird Count, indicated a possible increase, or at 
least stabilization, of goshawk numbers in the re- 
gion (Fig. 1). 

The status of Northern Goshawks is certainly 
tied to the distribution and condition of mature 
forest. However, the recent decline of some early- 
successional-stage species, such as grouse and lago- 
morphs (Rusch et al. 2000, Litvaitis 2001, Fuller 
and DeStefano 2003) , may influence goshawk dis- 
tribution and reproduction (Doyle and Smith 
1994) . Historically, the extirpation of some species, 
particularly the Passenger Pigeon, have likely al- 
tered the suite of available prey species for gos- 
hawks, while the expansion of some species, such 
as some passerines, in New England may provide 
new prey. Regardless, ubiquitous and intensive an- 
thropogenic change has characterized, and will 
continue to influence, the region’s landscape, veg- 
etation, and wildlife. DeGraaf and Yamasaki (2001: 


348 


DeStefano 


VoL. 39, No. 3 


3) summarized this by stating, “Most species have 
likely had very different distributions through 
time. In 50 or 100 years, both the species present 
and their distributions will be different.” This is 
likely the case for the Northern Goshawk. 

The characteristics of topography and forest cov- 
er reportedly used by goshawks in New England 
show similar patterns to other parts of the species’ 
range in North America. Nesting occurs in mature 
coniferous, deciduous, and mixed forest, typically 
on gentle rather than steep terrain, in proximity 
to some forest openings, but mostly away from well- 
used roads and human habitation. Similar patterns 
in nesting cover have been reported for other 
northeastern states (outside of New England) . In 
New Jersey and New York, goshawks selected ex- 
tensive mature forested areas for nesting, particu- 
larly in mixed hardwood-coniferous stands with 
greater numbers of large trees (>20 cm Diameter 
Breast Height) and high tree basal area (Allen 
1978, Speiser and Bosakowski 1987, Bosakowski 
and Speiser 1994). Hemlock, pine, and cedar {Cha- 
maecyparis thyoides) dominated nest sites, while oaks 
were less prevalent, although nests were usually in 
deciduous hardwood trees. Nests were present on 
gentle slopes or flat terrain, away from southern 
exposures, small forest tracts, paved roads, and hu- 
man habitation. 

Kenward (1996) speculated that goshawks in 
North America may face more competition from 
Red-tailed Hawks {Buteo jamaicensis) , Red-shoul- 
dered Hawks (B. lineatus) , and Great Horned Owls 
{Bubo vir^nianus) than goshawks face in Europe 
with similar raptor species. Red-tailed and Red- 
shouldered hawks are found throughout New En- 
gland, except for northernmost Maine for the Red- 
shouldered Hawk, and are regular breeders 
(DeGraaf and Yamasaki 2001). Red-shouldered 
Hawks inhabit mature deciduous-coniferous forest, 
while Red-tailed Hawks are found in more open 
habitats (DeGraaf and Yamasaki 2001). Great 
Horned Owls are uncommon, but widespread, and 
are found year-round throughout all of New En- 
gland, occurring in all types of cover (DeGraaf and 
Yamasaki 2001). Although little is known about the 
interactions among these raptor species, given the 
potential for aggressive interactions (Crannell and 
DeStefano 1992, Rohner and Doyle 1992), this may 
be an important local influence on the distribution 
of goshawks in some parts of the region. Broad- 
scale loss of hemlocks and the conditions of forest 
cover and canopy closure they create could also 


have important and related consequences for sev- 
eral wildlife species, including goshawks. 

New England: A “Natural Experiment” 

Keane and Morrison (1994), in the first sympo- 
sium on the ecology and management of Northern 
Goshawks, stressed the importance of identifying 
effects of scale and biological organization in eco- 
logical studies. In the same symposium, Graham et 
al. (1994) recommended that management of gos- 
hawks take place in large tracts of forest, which 
should be viewed as sustainable ecological units 
rather than smaller tracts or individual goshawk 
home ranges. Of the potential spatial scales that 
can be addressed, most forest wildlife biologists 
stressed the importance of, and need for, studies 
at large landscape levels (DeStefano 2002). 

New England would offer an interesting oppor- 
tunity to examine how goshawks have responded 
to a changing landscape. Widespread intensive 
land clearing and logging have given way to exten- 
sive reforestation of second- or multi-growth forest, 
embodying changes that have taken place over the 
last 2-3 centuries. Today, small, rare, and widely- 
dispersed patches of old-growth or virgin forest, al- 
tered disturbance regimes including reduced tim- 
ber harvest, dominant mid-aged forest, loss of 
early-successional-stage cover, and increases in hu- 
man densities and development offer an opportu- 
nity to see how goshawks have dealt with these 
changes in the northeast. This investigation would 
also offer insights into similar developing trends in 
the western U.S. Well distributed and coordinated 
monitoring of goshawk populations on randomly 
selected forested areas in New England, perhaps 
stratified by state, forest cover type, or ecological 
region, would be an appropriate approach. Surveys 
for goshawks could also include other forest rap- 
tors and m^or prey species, given recent emphasis 
away from single-species approaches and toward 
biodiversity (DeStefano 2002). However, the extent 
and effort required would be large, given the large 
spatial and temporal scales involved. 

Acknowledgments 

I am grateful to the biologists who took time out of 
their busy schedules to answer my questions about gos- 
hawks: Tom Hodgman (Maine), Scott Melvin (Massachu- 
setts), Christopher Gaughan (New Hampshire), Steve 
Parren, Steve Faccio, and Rosalind Renfrew (Vermont), 
Mariko Yamasaki (White Mountain National Forest), and 
Clayton Grove (Green Mountain National Forest). Bar- 
bara Loucks and Paul Novak also provided information 
on goshawks in neighboring New York. Special thanks to 


September 2005 


Conservation 


349 


Clint Boal for urging me forward on this topic, Mariko 

Yamasaki, Helen Snyder, and Michael Goldstein for re- 
viewing the manuscript, and Richard DeGraaf for our dis- 
cussions and his insights and knowledge of all things New 

England. 

Literature Cited 

Adamus, P.R. 1987. Atlas of breeding birds in Maine, 
1978—83. Maine Department of Inland Fisheries and 
Wildlife, Augusta, ME U.S.A. 

Allen, B.A. 1978. Nesting ecology of the goshawk in the 
Adirondacks. M.S. thesis, State Univ. New York Col- 
lege, Syracuse, NY U.S.A. 

Allen, G.M. 1903. A list of the birds of New Hampshire. 
Nature Study Press, Manchester, NH U.S.A. 

Andersen, D.E., S. DeStefano, M.I. Goldstein, K. Titus, 
D.C. Crocker-Bedford, JJ- Keane, R.G. Anthony, 
and R.N. Rosenfield. 2003. The status of the North- 
ern Goshawk in the western United States. Wildlife 
Society Technical Review 04-01, The Wildlife Society, 
Bethesda, MD U.S.A. 

Askins, R.A. 2001. Sustaining biological diversity in early 
successional communities: the challenge of managing 
unpopular habitats. Wildl. Soc. Bull. 29:407-412. 

Bagg, A.C. and S.A. Eliot, Jr. 1937. Birds of the Con- 
necticut Valley in Massachusetts. Hampshire Book- 
shop, Northampton, MA U.S.A. 

Bent, A.C. 1937. Life histories of North American birds 
of prey. Part I. U.S. National Museum Bulletin 170, 
Smithsonian Institution, Washington, DC U.S.A. 

Bevier, L.R. (Ed.). 1994. The adas of breeding birds of 
Connecticut. State Geological and Natural History 
Survey of Connecticut, Hartford, CT U.S.A. 

Birch, T.W. 1996. Private forest-land owners of the north- 
ern United States, 1994. U.S. Forest Service, Resource 
Bulletin NE-136, Radnor, PA U.S.A. 

Blockstein, D.E. 2002. Passenger Pigeon {Ectopistes mig- 
ratorius). In A. Poole and F. Gill, [Eds.], The birds of 
North America, No. 611. The Birds of North America, 
Inc., Philadelphia, PA U.S.A. 

Bosakowski, T. and R.E. Speiser. 1994. Macrohabitat se- 
lection by nesting Northern Goshawks: implications 
for managing eastern forests. Stud. Avian Biol. 16:46- 
49. 

Brooks, R.T. and T.W. Birch. 1988. Changes in New En- 
gland forests and forest owners: implications for wild- 
life habitat resources and management. Transactions 
of the North American Wildlife and Natural Resources Con- 
ference 53:78—87. 

Cogbill, C.V., J. Burk, and G. Motzkin. 2002. The for- 
ests of presetdement New England, USA: spatial and 
compositional patterns based on town proprietor sur- 
veys. /. Biogeography 29:1279-1304. 

Crannell, D. and S. DeStefano. 1992. An aggressive in- 
teraction between a Northern Goshawk and a Red- 
tailed Hawk./. Raptor Res. 26:269—270. 

Crocker-Bedford, D.C. 1998. The value of demographic 


and habitat studies in determining the status of 
Northern Goshawks {Accipiter gentilis atricapillus) with 
special reference to Crocker-Bedford (1990) and Ken- 
nedy (1997)./ Raptor Res. 32:329—336. 

Cronin, W. 1983. Changes in the land: Indians, colonists, 
and the ecology of New England. Hill and Wang, New 
York, NY U.S.A. 

Davis, M.B. 1996. Eastern old-growth forests, prospects 
for rediscovery and recovery. Island Press, Covelo, CA 
U.S.A. 

Dearborn, N. 1903. The birds of Durham and vicinity. 
Contributions from the Zoology Laboratory 6, New 
Hampshire College of Agriculture and the Mechanic 
Arts, Durham, NH U.S.A. 

DeGraaf, R.M. and M. Yamasaki. 2001. New England 
wildlife. Univ. Press of New England, Hanover, NH 
U.S.A. 

DeStefano, S. 1998. Determining the status of Northern 
Goshawks in the West: is our conceptual model cor- 
rect? /. Raptor Res. 32:342-348. 

. 2002. Regional and national issues for forest wild- 
life research and management. For. Sci. 48:181-189. 

Doyle, F.l. and J.M.N. Smith. 1994. Population response 
of Northern Goshawks to the 10-year cycle in num- 
bers of snowshoe hares. Stud. Avian Biol. 16:122-129. 

Ellsworth, J.W. and B.C. McComb. 2003. Potential ef- 
fects of Passenger Pigeon flocks on the structure and 
composition of presettlement forests of eastern North 
America. Conserv. Biol. 17:1548-1558. 

Enser, R.W. 1992. The atlas of breeding birds in Rhode 
Island. Rhode Island Department of Environmental 
Management, Providence, RI U.S.A. 

Forbush, E.H. 1925-1929. Birds of Massachusetts and 
other New England states. 3 vols. Massachusetts De- 
partment of Agriculture, Boston, MA U.S.A. 

Foss, C.R. 1994. Atlas of breeding birds in New Hamp- 
shire. Audubon Society of New Hampshire, Dover, 
NH U.S.A. 

Foster, D.R. 2002. Insights from historical geography to 
ecology and conservation: lessons from the New En- 
gland landscape./. Biogeography 29:1269—1275. 

, G. Motzkin, D. Bernardos, and J. Cardoza. 

2002. Wildlife dynamics in the changing New England 
landscape./. Biogeography 29:1337-1357. 

Fuller, T.K. and S. DeStefano. 2003. Relative impor- 
tance of early-successional forests and shrubland hab- 
itats to mammals in the northeastern United States. 
For. Ecol. Manage. 185:75—79. 

Graham, R.T., R.T. Reynolds, M.H. Reiser, R.L. Bassett, 
AND D.A. Boyce. 1994. Sustaining forest habitat for 
the Northern Goshawk: a question of scale. Stud. Avi- 
an Biol. 16:12—17. 

Griscom, L. and D.E. Snyder. 1955. The birds of Massa- 
chusetts, an annotated and revised check list. Peabody 
Museum, Salem, MA U.S.A. 

Hall, B., G. Motzkin, D.R. Foster, M. Syfert, and J. 
Burk. 2002. Three hundred years of forest and land- 


350 


DeStefano 


VoL. 39, No. 3 


use change in Massachusetts, U.S.A. J. Biogeogr. 29: 
1319-1335. 

Hoffman, R. 1904. A guide to the birds of New England 
and eastern New York. Houghton Mifflin, Boston, MA 
U.S.A. 

Keane, JJ- AND M.L. Morrison. 1994. Northern Goshawk 
ecology: effects of scale and levels of biological orga- 
nization. Stud. Avian Biol. 16:3-11. 

Kennedy, P.L. 1997. The Northern Goshawk {Accipiter gen- 
tilis atricapillus) : is there evidence of a population de- 
cline? J. Raptor Res. 31:95-106. 

Kenward, R.E. 1996. Goshawk adaptation to deforesta- 
tion: does Europe differ from North America? Pages 
233-243 in D. Bird, D. Varland, and J. Negro [Eds.], 
Raptors in hnman landscapes. Academic Press, Inc., 
San Diego, CA U.S.A. 

Kizlinski, M.L., D.A. Orwig, R.C. Cobb, and D.R. Foster. 
2002. Direct and indirect ecosystem consequences of 
an invasive pest on forests dominated by eastern hem- 
lock. /. 29:1489-1503. 

Laughlin, S.B. and D.P. Kibbe. [Eds.]. 1985. The atlas of 
breeding birds of Vermont. Univ. Press of New En- 
gland, Hanover, NH U.S.A. 

Litvaitis, J.A. 2001. Importance of early successional 
habitats to mammals in eastern forests. Wildl. Soc. Bull. 
29:466-473. 

Lorimer, C.G. 2001. Historical and ecological roles of 
disturbance in eastern North American forests: 9,000 
years of change. Wildl. Soc. Bull. 29:425-439. 

Orwig, D.A., D.R. Foster, and D.L. Mausel. 2002. Land- 
scape patterns of hemlock decline in New England 
due to the introduced hemlock woolly adelgid. J. Bio- 
geography 29:1475-1487. 

Paillet, el. 2002. Chestnut: history and ecology of a 
transformed species./. Biogeography 

Parker, G. 1995. Eastern coyote, the story of its success. 
Nimbus Publishing, Halifax, Nova Scotia, Canada. 

Parshall, T. and D.R. Foster. 2002. Fire on the New 
England landscape: regional and temporal variation, 
cultural and environmental controls. J. Biogeography 
29:1305-1317. 

Rohner, C. and R.I. Doyle. 1992. Food-stressed Great 


Horned Owl kills adult goshawk: exceptional obser- 
vation or community process? J. Raptor Res. 26:261— 
263. 

Rusch, D.H., S. DeStefano, M.G. Reynolds, and D. Lau- 
TEN. 2000. Ruffed Grouse {Bonasa umbellus). In A 
Poole and F. Gill [Eds.] The birds of North America, 
No. 515. The Birds of North America Inc., Philadel- 
phia, PA U.S.A. 

Sage, J.H., L.B. Bishop, and W.P. Bliss. 1913. The birds 
of Connecticut. Connecticut Geolo^cal and Nat. History 
Surv. Bull. 20. 

Sauer, J.R., J.E. Hines, andJ. Fallon. 2003. The North 
American breeding bird survey, results and analysis 
1966-2002. Version 2003.1, USGS Patuxent Wildlife 
Research Center, Laurel, MD U.S.A. (http://www. 
mp2-pwrc.usgs.gov/bbs/) 

Speiser, R. and T. Bosakowski. 1987. Nest site selection 
by Northern Goshawks in northern New Jersey and 
southeastern New York. Condor 89:387-394. 

Squires, J.R. and R.T. Reynolds. 1997. Northern Gos- 
hawk {Accipiter gentilis) . In A. Poole and F. Gill [Eds.], 
The birds of North America, No. 298. The Birds of 
North America Inc., Philadelphia, PA U.S.A. 

Thompson, F.R., III, and R.M. DeGraaf. 2001. Conser- 
vation approaches for woody, early successional com- 
munities in the eastern United States. Wildl. Soc. Bull. 
29:483-494. 

Tingley, M.W., D.A. Orwig, R. Field, and G. Motzkin 
2002. Avian response to removal of a forest dominant’ 
consequences of hemlock woolly adelgid infestations. 
J. Biogeography 29:1505-1516. 

Veit, R. and W.R. Petersen. 1993. Birds of Massachusetts. 
Mass. Audubon Society, Lincoln, MA U.S.A. 

Walker, L.C. 1999. The North American forests, geog- 
raphy, ecology, and silviculture. CRC Press, Boca Ra- 
ton, FL U.S.A. 

Wetherbee, D.K. 1945. The birds and mammals of 
Worcester County, Massachusetts. Century Press, 
Worcester, MA U.S.A. 

Received 24 March 2004; accepted 2 September 2004 

Associate Editor: Clint W. Boal 


A Telemetry Receiver Designed with 
The Researcher in Mind 


What you've been waiting for! 


Fmally. a highly sensitive 999 channel synthesized telemetry receiver that tweighs less than 13 ouncat, te 
completely user programmable and offers variable scan rates over all frequertdes. For each animal btkig 
tracked, the Urge LCD display provides not only the frequency (to lOOHa) and channel number, but also a 
7 character alphanumeric comment held and a digital signal strength meter. Stop carrying rtctivtrs that are 
the size of a lunch box or cost over $1500. The features and performance of the new R-1000 podcet sized 
telemetry receiver will impress you. and the price will convince you. 

Other Ftaturts include: 

_ • Factory tuned to any 4HHz wide segment in the 144174MHz Band • Very high sensitivity of -liBdBm to 
ISOdBm • Illuminated display and keypad for use in tow light or darkness • User selectable Kan rates 
from 1-30 seconds in 1 second steps • Rechargeable battenes operate the receiver for 12 hours and 

can be repUced with standaH AA Alkaline batteries in the held. 
Both 12vdc and llOvac chargers are included. 

% 






• 6.r (15.5cm) high. 

2.6" (6.6cm) wide. 

1.5' (3.8cm) deep. 

• 3 year warranty 

• 1 day delivery 

$ 695.00 

Please specify desired 4MHz 
wide segment in the 
148- 174MHz band 


Visit our 
website for 
complete 
specihcations. 
operating 
manual and 
information 
on the R-lOOO 
or call our 
toll-free number 
to order your 
receiver now. 

Try tht 
New R-1000 
and YouV 
Imprtssadl 






COMMUNICATIONS SPECIALISTS. INC. 


I.iM A. • ('-f.i'Mjc, (A • : M-. 'i'J8 lO.-’l • f.i- 1 

E'itir.* U.S.M_ (H0\') Hv.. • F.it iHH'M mSO Ov.*; • tilt;); a a.l i -n'. 




2006 ANNUAL MEETING 


The Raptor Research Foundation, Inc. 2006 annual meeting will be held in conjunction with the 
Fourth North American Ornithological Conference on 3-7 October 2006 in Veracruz, Mexico. For 
more information about the meeting see http://www.naoc2006.org/ 

Persons interested in predatory birds are invited to join The Raptor Research Foundation, Inc. (see: 
http:/ /biology.boisestate.edu/raptor/). Send requests for information concerning membership, subscriptions, 
special publications, or change of address to OSNA, 5400 Bosque Blvd., Suite 680, Waco TX 76710, U.S.A. 

The Journal of Raptor Research (ISSN 0892-1016) is published quarterly and available to individuals for $40.00 
per year and to libraries and institutions for $65.00 per year from The Raptor Research Foundation, Inc., 14377 
1 l7th Street South, Hastings, Minnesota 55033, U.S.A. (Add $3 for destinations outside of the continental United 
States.) Periodicals postage paid at Hastings, Minnesota, and additional mailing offices. POSTMASTER: Send 
address changes to The Journal of Raptor Research, OSNA, P.O. Box 1897, Lawrence, KS 66044-8897, U.S.A. 

Printed by Allen Press, Inc., Lawrence, Kansas, U.S.A. 

Copyright 2005 by The Raptor Research Foundation, Inc. Printed in U.S.A. 

© This paper meets the requirements of ANSi/NiSO Z39.48-1992 {Permanence of Paper). 


Raptor Research Foundation, Inc. 

Grants and Awards 

For details and additional information visit: http:/ /biology.boisestate.edu/raptor/grantsandawards.htm 

Awards for Recognition of Significant Contributions 

The Tom Cade Award is a non-monetary award that recognizes an individual who has made significant advances 
in the area of captive propagation and reintroduction of raptors. The Fran and Frederick Hamerstrom 
Award is a non-monetary award that recognizes an individual who has contributed significantly to the under- 
standing of raptor ecology and natural history. Submit nominations for either award to: Dr. Clint Boal, Texas 
Cooperative Fish and Wildlife Research Unit, BRD/USGS, Texas Tech University, 15th Street & Boston, 
Ag Science Bldg., Room 218, Lubbock TX 79409-2120 U.S.A.; phone: 806-742-2851; e-mail: cboal@ttu.edu 

Awards for Student Recognition and Travel Assistance 

The James R. Koplin Travel Award is given to a student who is the senior author and presenter of a paper or 
poster to be presented at the RRF meeting for which travel funds are requested. Application deadline: due 
date for meeting abstract. Contact: Dr. Patricia A. Hall, 5937 E. Abbey Rd., Flagstaff, AZ 86004; phone: 
520-526-6222 U.S.A.; e-mail: pah@spruce.for.nau.edu 

The William C. Anderson Memorial Award is given to both the best student oral and poster presentation at the 
annual RRF meeting. The paper cannot be part of an organized symposium to be considered. Application 
deadline: due date for meeting abstract, no special application is needed. Contact: Rick Gerhardt, Sage 
Science, 319 SE Woodside Ct., Madras, OR 97741 U.S.A; phone: 541-475-4330; email: rgerhardt@madras.net 

Grants 

Application deadline for all grants is February 15 of each year; selections will be made by April 15. 

The Dean Amadon Grant for up to $1000 is designed to assist persons working in the area of systematics (tax- 
onomy) and distribution of raptors. The Stephen R. TuUy Memorial Grant for up to $500 is given to sup- 
port research and conservation of raptors, especially to students and amateurs with limited access to alter- 
native funding. Agency proposals are not accepted. Contact for both grants: Dr. Carole Griffiths, 251 
Martling Ave., Tarrytown, NY 10591 U.S.A.; phone: 914-631-2911; e-mail: cgriff@liu.edu 

The Leslie Brown Memorial Grant for up to $1400 is given to support research and/or the dissemination of 
information on African raptors. Contact: Dr. Jeffrey L. Lincer, 9251 Golondrina Drive, La Mesa, CA 91941, 
U.S.A.; e-mail: JeffLincer@tns.net